feat(gitops): add isolated detached worktree manager

docs(aios-v2): add CTO implementation status and gap analysis
create branch for AIOS V2 implementation status documentation
2026-07-20 00:42:56 +08:00 · 2026-07-20 00:25:57 +08:00 · 2026-07-20 00:21:00 +08:00 · 2026-07-14 01:33:53 +08:00 · 2026-07-12 21:41:10 +08:00 · 2026-07-12 19:58:26 +08:00
238 changed files with 16264 additions and 137 deletions
@@ -0,0 +1,156 @@
+# V2R1 Gate 1 — Clean-Room Final Evidence
+
+## Summary
+
+| Field | Value |
+|---|---|
+| Date | 2026-07-14 |
+| Source commit | `502ef30642597fb0b70b90ae2e8a0b6af217d80e` |
+| Source branch | `origin/work/v2r1-integration-quality-2000` |
+| Result branch | `fix/v2r1-g1-cleanroom-final-evidence-r2-20260714` |
+| Result commit reference | `HEAD` |
+| Gate | `G1` |
+| Variant | `openclaw-cleanroom` |
+| Status | `PASS` |
+
+The immutable report uses `HEAD` for post-commit verification. The resolved commit object ID is recorded by the external acceptance result after this file is committed; a Git commit cannot reliably embed its own object ID because changing the file changes the tree and commit identity.
+
+## Machine-Readable Result
+
+```yaml
+gate: G1
+variant: openclaw-cleanroom
+date: 2026-07-14
+status: PASS
+source_commit: 502ef30642597fb0b70b90ae2e8a0b6af217d80e
+source_branch: origin/work/v2r1-integration-quality-2000
+result_branch: fix/v2r1-g1-cleanroom-final-evidence-r2-20260714
+result_commit_ref: HEAD
+tests_total: 32
+tests_passed: 32
+tests_failed: 0
+cleanup_scan: PASS
+secret_scan: PASS
+rollback_status: PASS
+recovery_verified: true
+workspace_clean_at_test_completion: true
+merge_performed: false
+tag_created: false
+baseline_promoted: false
+deployment_performed: false
+```
+
+## Evidence Provenance
+
+The test, cleanup, secret-scan, rollback, and recovery results below are execution evidence supplied by OpenClaw for the Gate 1 run. This correction commit changes only the evidence report and does not claim that the GitOps API reran the suites.
+
+## HMAC Test-Key Length Verification
+
+| Test file | Variable | Test-only value | Length |
+|---|---|---|---:|
+| `tests/v2r1/test_runtime_wiring.py:24` | `internal_key` | `b"runtime-internal-context-key-is-32bytes"` | 39 bytes |
+| `tests/v2r1/test_runtime_wiring.py:25` | `ingress_key` | `b"runtime-external-ingress-key-is-32bytes"` | 39 bytes |
+
+Both values are test fixtures. The supplied evidence reports that no production secret was introduced.
+
+## Test Results Supplied by OpenClaw
+
+### V2R1 suite — 29 of 29 passed
+
+```shell
+PYTHONPATH=active-runtime/app python3 -m pytest \
+  tests/v2r1/test_hermes_fabric.py \
+  tests/v2r1/test_message_ledger_append.py \
+  tests/v2r1/test_router_isolation.py \
+  tests/v2r1/test_runtime_identity_binding.py \
+  tests/v2r1/test_runtime_wiring.py \
+  tests/v2r1/test_tenant_context.py -v
+# Reported result: 29 passed, exit code 0
+```
+
+### Integration suite — 3 of 3 passed
+
+```shell
+PYTHONPATH=active-runtime/app python3 -m pytest tests/integration/test_runtime_ingress_flow.py -v
+# Reported result: 3 passed, exit code 0
+```
+
+**Reported total: 32 of 32 passed.**
+
+## Cleanup, Security, and Recovery Evidence
+
+| Check | Supplied result |
+|---|---|
+| Conflict-marker scan | PASS |
+| Cleanup scan | PASS |
+| Secret scan | PASS |
+| Legitimate hardcoded test keys | Two 39-byte fixtures in `test_runtime_wiring.py` |
+| Rollback rehearsal | PASS; 13 of 13 core tests reported after rollback to `6077331` |
+| Recovery verification | PASS |
+
+## Deployment State
+
+| Operation | Performed |
+|---|---|
+| Merge | No |
+| Tag creation | No |
+| Baseline promotion | No |
+| Staging deployment | No |
+| Production deployment | No |
+
+## Post-Commit Verification
+
+Run from the repository root after the commit is created:
+
+```shell
+set -eu
+
+SOURCE=502ef30642597fb0b70b90ae2e8a0b6af217d80e
+RESULT_FILE=.agent/results/openclaw/RESULT-V2R1-G1-OPENCLAW-FINAL-EVIDENCE-20260714.md
+
+git rev-parse HEAD
+
+test "$(git diff --name-only "$SOURCE" HEAD | wc -l | tr -d ' ')" = "1"
+test "$(git diff --name-only "$SOURCE" HEAD)" = "$RESULT_FILE"
+
+git ls-tree -r HEAD -- "$RESULT_FILE"
+
+python3 - "$RESULT_FILE" <<'PY'
+from pathlib import Path
+import re
+import sys
+
+path = Path(sys.argv[1])
+text = path.read_text(encoding="utf-8")
+
+encoded_tokens = [
+    [84, 66, 68],
+    [80, 69, 78, 68, 73, 78, 71],
+    [84, 79, 68, 79],
+    [70, 73, 88, 77, 69],
+    [91, 123, 99, 111, 109, 109, 105, 116, 125, 93],
+]
+tokens = ["".join(chr(value) for value in item) for item in encoded_tokens]
+angle_token = re.compile(chr(60) + r"[^>\n]+" + chr(62))
+
+hits = [token for token in tokens if token in text]
+if angle_token.search(text):
+    hits.append("angle-delimited-token")
+
+if hits:
+    print("TOKEN_SCAN=FAIL", hits)
+    raise SystemExit(1)
+
+print("TOKEN_SCAN=PASS")
+PY
+```
+
+## Acceptance Conditions
+
+Gate 1 is eligible to proceed only when the external acceptance record confirms all of the following against the resolved `HEAD` commit:
+
+- the commit exists and has source commit `502ef30642597fb0b70b90ae2e8a0b6af217d80e` as its parent;
+- exactly this one report file changed;
+- the committed report contains no unresolved legacy token;
+- the result tree is not identical to rejected commit `71239b4d187e683e05fd22fc9d5df8ae324e3390`;
+- no merge, tag, baseline promotion, or deployment occurred.
@@ -10,3 +10,4 @@ node_modules/
 runtime/tmp/
 runtime/cache/
 logs/
+config/test_write_perms
@@ -0,0 +1,218 @@
+# AIOS V2 Implementation Status
+
+- 扫描日期：2026-07-20
+- 当前发布基线：`V1R4C6`
+- 基线提交：`e204a90398bf646d46245b3088226749fa056edd`
+- 实施分支：`patch/aios-v2-implementation-status-next-phase-20260720`
+- 决策角色：AIOS V2 CTO
+- 总体判定：**进入主线收敛阶段，不重写现有系统。**
+
+## 1. 执行摘要
+
+AIOS V2 已经具备一批可复用的候选实现，包括签名租户上下文、权威 Tenant Router、Tenant Cell 生命周期、OpenClaw Fleet 适配、Hermes 租户队列与 Profile 隔离、消息账本、内存/密钥作用域以及追加式审计链。GitOps Runtime 当前可用，分支、提交和审计元数据机制可以继续作为所有变更的唯一入口。
+
+当前尚不能宣称 AIOS V2 已交付。`V1R4C6` 的提升理由仍注明未部署生产；V2R1 实现主要存在于候选发布、集成、任务和结果分支，未进入 `main` 或当前基线。请求的 `docs/aios-v2/` 权威设计目录在已核对的 `main`、当前治理分支和抽查的 V2R1 集成/设计分支中未发现，因此架构目标尚未形成可被自动验证的单一事实源。
+
+CTO 决策：
+
+1. 不推倒 OpenClaw、Hermes、Agent Router 和现有 GitOps 能力。
+2. AIOS 收敛为唯一控制面；OpenClaw 变为执行适配器；Hermes 变为租户隔离的 Runtime；Agent Router 变为统一调度与路由平面。
+3. 候选能力先做证据映射、边界硬化和小步集成，再进入基线。
+4. 所有 API、控制能力、运行状态与审计查询必须 Web 化。
+5. 所有模块必须在入口、存储、队列、缓存、密钥、日志、审计和前端会话层显式携带并校验 `tenant_id`。
+6. 所有修改必须通过非 `main` 分支、可审查提交、自动测试和受控提升完成。
+
+## 2. 扫描范围与证据边界
+
+### 已读取
+
+- GitOps Runtime 状态与当前基线。
+- `main`、V2R1 候选发布分支、集成/任务/结果分支的定向差异。
+- `release/v2r1c0-tenant-cell-fabric-20260712` 中的 `active-runtime/app/aios_v2r1` 候选实现。
+- 同一候选分支中的 `tests/v2r1` 合同测试与发布门禁说明。
+- Gateway、自动编排、租户隔离、WebUI、事件总线等相关分支和结果分支的提交证据。
+
+### 限制
+
+- `docs/aios-v2/` 未出现在已核对的权威分支，无法完成“对该目录全部文件逐份比对”。本报告将此项列为 P0，而不是假设文档已经存在。
+- 从 `V1R4C6` 到 `main`、以及部分大型任务分支的累计差异超过 GitOps diff 接口上限。因此本次采用定向目录、最近提交和结果证据扫描；下一阶段必须生成机器可读的能力—提交—测试索引。
+- “分支存在”不等于“能力已合入”；“测试报告存在”不等于“生产门禁已通过”。状态必须区分基线、候选、验证和部署。
+
+## 3. 当前架构扫描
+
+### 3.1 权威基线
+
+| 项目 | 当前状态 | 结论 |
+|---|---|---|
+| 发布基线 | `V1R4C6` | 当前唯一被提升的版本 |
+| 生产部署 | 未确认；基线理由注明未部署生产 | 不得视为生产完成 |
+| GitOps Runtime | Active | 可作为唯一变更通道 |
+| V2R1 代码进入 `main` | 未发现 | 候选能力尚未成为权威实现 |
+| `docs/aios-v2/` | 未发现 | P0 文档与治理缺口 |
+
+### 3.2 候选运行时能力
+
+`release/v2r1c0-tenant-cell-fabric-20260712` 中已发现以下候选模块：
+
+| 模块 | 已有能力 | 当前成熟度 |
+|---|---|---|
+| `tenant_context.py` | HMAC 签名上下文、规范化消息模型、来源校验、基础工具策略 | 候选；缺少密钥轮换、重放防护和集中策略版本 |
+| `router.py` | SQLite 权威租户、渠道、主体和会话路由；fail-closed 校验 | 候选；需统一到控制面并替换单机存储假设 |
+| `cell_manager.py` / `cli.py` | Tenant Cell 注册和状态操作 | 原型；JSON 注册表无并发控制，状态机约束不足 |
+| `fleet.py` | OpenClaw Fleet 命令适配 | 可复用适配层雏形；缺少超时、认证、幂等、审计和能力协商 |
+| `hermes_fabric.py` | 强制租户分区的任务队列、来源限制、领取接口 | 候选；共享 SQLite 仍是逻辑隔离，需生产级队列和配额 |
+| `hermes_profile.py` | HOME、工作区和路径逃逸防护 | 可复用；需容器/进程级隔离证据 |
+| `memory_scope.py` | session/profile/cell/tenant/shared artifact 命名空间 | 可复用合同；需绑定实际存储后端和查询层 |
+| `secret_scope.py` | 仅保存 SecretRef，校验 tenant/cell 所有权 | 可复用合同；尚无真实 broker、轮换和撤销流程 |
+| `message_ledger.py` | 精确消息 ID、TTL、冲突检测和引用解析 | 候选；需迁移、容量、加密、备份和多实例并发验证 |
+| `audit.py` | 追加式哈希链 | 原型；当前链为全局序列，应按租户/审计域隔离并支持外部不可变存储 |
+
+### 3.3 测试与门禁
+
+已发现可执行合同测试，覆盖 Tenant Router 隔离、TenantContext 签名和 Hermes 来源/重复事件处理。候选验收清单明确要求：
+
+- 跨租户内存、Cell、密钥和队列访问必须拒绝。
+- 非当前已认证人工命令不得触发副作用。
+- OpenClaw Runtime Hook 与副作用审批属于必需项。
+- 在可执行测试与运行时证据通过之前，不得生产部署。
+
+因此当前标记为“候选能力存在”，而不是“发布完成”。
+
+### 3.4 控制面与自动化
+
+结果分支显示 Gateway 和自动编排已出现较强测试证据，包括单元/集成测试与 Codex 任务领取、生命周期回写能力。但是这些成果仍分散在结果/集成分支，尚未形成 AIOS 唯一入口，也未完成从任务结果到权威基线的提升闭环。
+
+### 3.5 Web 化现状
+
+仓库存在 WebUI、观测与控制面相关分支，但分支名称和最后提交内容有不一致现象。当前不能仅依据分支名认定功能完成。需要建立 API 路由、页面、权限、租户隔离、E2E 测试和部署证据五项映射。
+
+## 4. 与 AIOS V2 目标架构对比
+
+目标逻辑架构：
+
+```text
+Web Console / Public API
+          |
+AIOS Control Plane
+  - Tenant Registry
+  - Identity & Policy
+  - Agent Registry
+  - Workflow / Task Lifecycle
+  - Approval & Audit
+  - Runtime Inventory
+          |
+Agent Router / Scheduling Plane
+       /              \
+OpenClaw Adapter    Hermes Runtime Adapter
+       |              |
+OpenClaw Cells      Hermes Workers / Profiles
+```
+
+边界规则：
+
+- 只有 AIOS Control Plane 可以创建、审批、取消和审计任务。
+- Router 只负责基于权威上下文的选择、调度和转发，不拥有租户事实源。
+- OpenClaw 不拥有租户、策略或任务生命周期；只执行有签名、可审计、可幂等的命令。
+- Hermes 不拥有控制面策略；只运行隔离 Profile、消费租户队列并返回结果/遥测。
+- Web Console 不直接调用 OpenClaw/Hermes；只调用 AIOS API。
+- 每次调用均传播 `tenant_id`、`principal_id`、`session_id`、`task_id`、`trace_id` 和策略版本。
+
+## 5. Gap Analysis
+
+| ID | 领域 | 当前 | 目标 | Gap | 优先级 |
+|---|---|---|---|---|---|
+| G-001 | 架构事实源 | `docs/aios-v2/` 不在权威分支 | 全部 V2 设计、ADR、接口和门禁可追踪 | 无法自动核对目标架构 | P0 |
+| G-002 | 主线收敛 | V2 能力分散于候选/结果分支 | `main` 上单一集成历史 | 分支存在但未纳管、未提升 | P0 |
+| G-003 | 唯一控制面 | Gateway/Router/执行器均可能承载控制逻辑 | AIOS 是唯一任务与策略入口 | 控制权边界尚未强制 | P0 |
+| G-004 | 多租户强制 | 候选模块多为显式字段和 SQLite 逻辑隔离 | 全链路强制隔离、负向测试、生产存储策略 | 缺少统一 middleware、RLS/物理隔离策略和 E2E 证据 | P0 |
+| G-005 | 审计与重放 | 全局哈希链；上下文有 nonce 但无 replay store | 按租户/审计域不可篡改、可查询、可验证；命令防重放 | 审计域、保留、导出、重放防护不足 | P0 |
+| G-006 | OpenClaw 边界 | Fleet CLI 适配已有 | 版本化执行协议、幂等、超时、取消、回调、审计 | 仍可能被控制面之外直接调用 | P1 |
+| G-007 | Hermes 边界 | 租户队列/Profile 已有 | Runtime-only、配额、公平调度、取消、恢复 | 共享存储和控制逻辑仍需拆分 | P1 |
+| G-008 | Web 化 | 多个局部 Web/UI 分支 | 所有控制能力均有 Web/API，权限一致 | 无统一 Console、API contract 和 E2E 门禁 | P1 |
+| G-009 | 生产数据层 | 多个候选模块使用 SQLite/JSON | HA、迁移、备份、加密、容量和恢复目标 | 原型数据层不能直接生产 | P1 |
+| G-010 | 证据治理 | 分支名与内容可能不一致，大 diff 超限 | 能力—提交—测试—部署机器可读映射 | 无法可靠计算完成度 | P0 |
+| G-011 | 发布闭环 | 基线提升与生产部署脱节 | 构建、测试、安全、审批、提升、部署、回滚闭环 | 当前基线仍注明未部署 | P0 |
+| G-012 | 可观测性 | 局部分支存在观测/WebUI 工作 | tenant/task/agent/trace 统一指标日志追踪 | 统一 telemetry contract 尚未成为门禁 | P1 |
+
+## 6. 能力标记
+
+### 6.1 保留并复用
+
+- GitOps Runtime、非主线分支、提交元数据、审批式 Tag/回滚流程。
+- OpenClaw Fleet 和现有执行能力。
+- Hermes Runtime、Profile、队列和工作区能力。
+- Agent Router 的路由/调度经验和已有接口。
+- V2R1 TenantContext、CanonicalMessage、TenantRouter、MemoryScope、SecretRef、MessageLedger 合同。
+- 现有 Gateway、自动领取、任务生命周期回写和测试资产。
+- 已有 WebUI、审计和观测代码中可以由证据证明的部分。
+
+### 6.2 必须重构而非重写
+
+- 将 OpenClaw 内部的租户、策略、审批和任务状态决策移到 AIOS；保留执行器和 Fleet 能力。
+- 将 Hermes 的控制逻辑移到 AIOS/Router；保留 Runtime、Profile 与 Worker。
+- 合并重复 Router/Gateway 入口，形成一个权威 Tenant/Policy middleware。
+- 将 SQLite/JSON 原型抽象为 repository/queue/audit 接口，允许先兼容、后替换生产后端。
+- 将审计链改为租户或明确审计域分区，并加入签名、外部固化、查询授权和验证作业。
+- 将所有 UI 改为 AIOS API 客户端，禁止浏览器直接访问执行层。
+- 将测试报告升级为可复现命令、commit、exit code、pass count 和 artifact digest。
+
+### 6.3 暂停继续扩展
+
+在 P0 收敛完成前，不新增：
+
+- 绕过 AIOS Gateway 的 OpenClaw/Hermes 公共入口。
+- 未携带 TenantContext 的任务、消息、缓存或后台作业。
+- 只存在于个人/临时分支且没有验收合同的大型功能。
+- 单次超大提交；每个任务必须可独立审查、测试和回滚。
+
+## 7. 实施路线
+
+### Phase 0：权威化与门禁（立即）
+
+1. 创建 `docs/aios-v2/` 单一事实源和 ADR 索引。
+2. 生成能力—提交—测试—部署 Evidence Index。
+3. 提取 V2R1 候选能力到窄范围集成分支，不直接合并整个历史分支。
+4. 建立统一 TenantContext middleware、Policy Decision 和审计接口。
+5. 建立 AIOS Control Plane Gateway 唯一入口与禁止旁路测试。
+6. 建立 CI Gate：多租户负向测试、重放、幂等、权限、Web E2E、审计验证。
+
+### Phase 1：执行层解耦
+
+1. OpenClaw Adapter v1：版本化协议、能力发现、start/stop/execute/cancel/status、幂等键、超时、结果回调。
+2. Hermes Runtime Adapter v1：enqueue/claim/heartbeat/cancel/result、配额、公平调度和 Tenant Profile 强制。
+3. Router 只读取 AIOS 权威注册表与策略结果，不保留第二套租户事实源。
+
+### Phase 2：Web 与生产化
+
+1. AIOS Console：Tenant、Agent、Workflow、Task、Approval、Runtime、Audit、GitOps 页面。
+2. 统一 API、RBAC/ABAC、租户会话隔离、CSRF/CORS、速率限制和操作审计。
+3. 替换/封装原型数据层，完成迁移、备份恢复、HA、容量和灾备演练。
+4. 完成受控基线提升、部署、冒烟和回滚证据。
+
+## 8. 下一阶段出口标准
+
+只有同时满足以下条件，才允许进入 V2R1 集成候选：
+
+- `docs/aios-v2/` 在目标集成分支存在，且架构、ADR、API、租户模型和门禁互相链接。
+- AIOS Gateway 是任务创建、审批、状态变更和取消的唯一公共入口。
+- OpenClaw/Hermes 旁路调用被代码、网络策略或测试明确拒绝。
+- 所有持久化记录和消息均含不可为空的 `tenant_id`，跨租户负向测试通过。
+- TenantContext 具备密钥版本、重放缓存、时钟偏差、撤销和审计策略。
+- 审计按租户/审计域隔离，能够从 Web/API 授权查询并验证完整性。
+- Codex 和 OpenClaw 任务均返回 branch、commit、测试命令、exit code、pass count、风险与回滚说明。
+- 变更拆分为可审查提交，不出现无法由 diff 接口检查的单任务超大提交。
+- 未执行直接 `main` push、Tag、生产部署或密钥变更。
+
+## 9. 当前状态结论
+
+| 结论 | 状态 |
+|---|---|
+| 不重写现有代码 | 已决策 |
+| 优先复用 OpenClaw/Hermes/Agent Router | 已决策 |
+| OpenClaw 与 Hermes 逐步解耦 | 已列入 Phase 1 |
+| AIOS 成为唯一控制面 | P0 实施中 |
+| 所有功能 Web 化 | P1 规划完成，尚未验收 |
+| 所有模块多租户隔离 | 候选合同存在，生产级全链路尚未完成 |
+| 所有修改通过 Git 任务流程 | 当前已遵守 |
+
+**总体完成度判定：架构与候选能力约处于“可收敛、不可发布”阶段。下一步不是继续增加功能，而是建立权威事实源、统一入口、租户门禁和可验证的集成闭环。**
@@ -1 +1 @@
-V1R1C4
+V1R4C6
@@ -0,0 +1,56 @@
+"""API adapter for agent context routing and visibility."""
+
+from typing import Any
+
+try:
+    from fastapi import APIRouter
+except Exception:
+    APIRouter = None
+
+from .agent_context_dashboard import AgentContextDashboard
+from .agent_context_rbac import AgentContextRBAC
+from .agent_context_resolver import AgentContextResolver
+
+resolver = AgentContextResolver()
+rbac = AgentContextRBAC()
+dashboard = AgentContextDashboard()
+
+
+def resolve_agent_context(payload: dict[str, Any]) -> dict[str, Any]:
+    route = rbac.route(payload["user_id"], payload.get("group_id"), payload.get("role", "user"))
+    context = resolver.resolve(
+        user_id=payload["user_id"],
+        group_id=payload.get("group_id", ""),
+        session=payload.get("session"),
+        user_profile=payload.get("user_profile"),
+        group_policy=payload.get("group_policy"),
+        agent_default=payload.get("agent_default"),
+        owner_policy=payload.get("owner_policy"),
+        owner=payload.get("role") == "owner",
+    )
+    return {"routing": route.to_dict(), "context": context.to_dict()}
+
+
+def get_context_dashboard() -> dict[str, Any]:
+    return dashboard.overview()
+
+
+def build_router():
+    if APIRouter is None:
+        return None
+    router = APIRouter(prefix="/admin/ai-evolution/context", tags=["context-isolation"])
+
+    @router.post("/resolve", operation_id="resolveAgentContext")
+    def resolve(payload: dict[str, Any]):
+        return resolve_agent_context(payload)
+
+    @router.get("/overview", operation_id="getAgentContextDashboard")
+    def overview():
+        return get_context_dashboard()
+
+    return router
+
+
+router = build_router()
+
+__all__ = ["router", "resolve_agent_context", "get_context_dashboard"]
@@ -0,0 +1,23 @@
+"""Audit context resolution decisions."""
+
+from datetime import datetime, timezone
+from typing import Any
+
+
+def utc_now():
+    return datetime.now(timezone.utc).isoformat()
+
+
+class AgentContextAudit:
+    def record(self, context: dict[str, Any]) -> dict[str, Any]:
+        return {
+            "event": "context_loaded",
+            "agent_id": context.get("agent_id"),
+            "user_id": context.get("user_id"),
+            "group_id": context.get("group_id"),
+            "layers": [x.get("name") for x in context.get("layers", [])],
+            "timestamp": utc_now(),
+        }
+
+
+__all__ = ["AgentContextAudit"]
@@ -0,0 +1,47 @@
+"""Web UI model for context isolation visibility."""
+
+from __future__ import annotations
+
+from typing import Any
+
+
+class AgentContextDashboard:
+    def overview(self, contexts: list[dict[str, Any]] | None = None) -> dict[str, Any]:
+        contexts = contexts or []
+        return {
+            "module": "agent-context-isolation",
+            "route": "/admin/ai-evolution/context",
+            "title": "Agent Context Isolation",
+            "panels": [
+                {
+                    "name": "agent-routing",
+                    "title": "Agent 路由",
+                    "data": contexts,
+                },
+                {
+                    "name": "context-layers",
+                    "title": "Context 来源层级",
+                    "layers": [
+                        "safety_policy",
+                        "owner_policy",
+                        "group_policy",
+                        "role_policy",
+                        "user_profile",
+                        "session_context",
+                        "agent_default",
+                    ],
+                },
+                {
+                    "name": "audit",
+                    "title": "Context 审计",
+                    "events": [
+                        "context_loaded",
+                        "profile_updated",
+                        "agent_selected",
+                    ],
+                },
+            ],
+        }
+
+
+__all__ = ["AgentContextDashboard"]
@@ -0,0 +1,55 @@
+"""RBAC binding for group main agents and owner agents."""
+
+from __future__ import annotations
+
+from dataclasses import asdict, dataclass
+from typing import Any
+
+
+@dataclass(frozen=True)
+class AgentRoutingDecision:
+    user_id: str
+    group_id: str
+    role: str
+    agent_id: str
+    scope: str
+    allowed: bool
+    reason: str
+
+    def to_dict(self) -> dict[str, Any]:
+        return asdict(self)
+
+
+class AgentContextRBAC:
+    """Routes users to group agents and owners to dedicated owner agents."""
+
+    VALID_ROLES = {"user", "group_admin", "owner"}
+
+    def route(self, user_id: str, group_id: str | None, role: str) -> AgentRoutingDecision:
+        if role not in self.VALID_ROLES:
+            return AgentRoutingDecision(user_id, group_id or "", role, "", "none", False, "unknown_role")
+        if role == "owner":
+            return AgentRoutingDecision(
+                user_id=user_id,
+                group_id=group_id or "global",
+                role=role,
+                agent_id=f"owner-agent:{user_id}",
+                scope="global",
+                allowed=True,
+                reason="owner_dedicated_agent",
+            )
+        if not group_id:
+            return AgentRoutingDecision(user_id, "", role, "", "none", False, "group_required")
+        scope = "group" if role == "group_admin" else "self"
+        return AgentRoutingDecision(
+            user_id=user_id,
+            group_id=group_id,
+            role=role,
+            agent_id=f"group-main-agent:{group_id}",
+            scope=scope,
+            allowed=True,
+            reason="group_agent_routing",
+        )
+
+
+__all__ = ["AgentRoutingDecision", "AgentContextRBAC"]
@@ -0,0 +1,82 @@
+"""Multi-user agent context resolver.
+
+Architecture:
+- shared agent kernel
+- group main agent
+- owner dedicated agent
+- user profile overlay
+- session overlay
+
+Personal preferences never enter shared USER.md.
+"""
+
+from __future__ import annotations
+
+from dataclasses import asdict, dataclass, field
+from datetime import datetime, timezone
+from typing import Any
+
+
+def utc_now() -> str:
+    return datetime.now(timezone.utc).isoformat()
+
+
+@dataclass(frozen=True)
+class ContextLayer:
+    name: str
+    priority: int
+    data: dict[str, Any]
+
+
+@dataclass
+class EffectiveContext:
+    agent_id: str
+    user_id: str
+    group_id: str
+    layers: list[ContextLayer]
+    resolved: dict[str, Any]
+    resolved_at: str = field(default_factory=utc_now)
+
+    def to_dict(self):
+        return {
+            "agent_id": self.agent_id,
+            "user_id": self.user_id,
+            "group_id": self.group_id,
+            "layers": [asdict(x) for x in self.layers],
+            "resolved": self.resolved,
+            "resolved_at": self.resolved_at,
+        }
+
+
+class AgentContextResolver:
+    """Resolves context without shared user-state pollution."""
+
+    def resolve(
+        self,
+        user_id: str,
+        group_id: str,
+        session: dict[str, Any] | None = None,
+        user_profile: dict[str, Any] | None = None,
+        group_policy: dict[str, Any] | None = None,
+        agent_default: dict[str, Any] | None = None,
+        owner_policy: dict[str, Any] | None = None,
+        owner: bool = False,
+    ) -> EffectiveContext:
+        layers = [
+            ContextLayer("agent_default", 10, agent_default or {}),
+            ContextLayer("group_policy", 30, group_policy or {}),
+            ContextLayer("user_profile", 50, user_profile or {}),
+            ContextLayer("session", 70, session or {}),
+        ]
+        agent_id = f"group-main-agent:{group_id}"
+        if owner:
+            agent_id = f"owner-agent:{user_id}"
+            layers.append(ContextLayer("owner_policy", 90, owner_policy or {}))
+        resolved: dict[str, Any] = {}
+        # Higher priority overlays lower priority except safety policies handled upstream.
+        for layer in sorted(layers, key=lambda x: x.priority):
+            resolved.update(layer.data)
+        return EffectiveContext(agent_id, user_id, group_id, layers, resolved)
+
+
+__all__ = ["ContextLayer", "EffectiveContext", "AgentContextResolver"]
@@ -0,0 +1,77 @@
+"""Actions API for unified Agent Event Bus."""
+
+from __future__ import annotations
+
+from typing import Any
+
+try:
+    from fastapi import APIRouter
+except Exception:
+    APIRouter = None
+
+from .agent_event_bus import AgentEventBus
+
+bus = AgentEventBus()
+
+
+def create_agent_event(payload: dict[str, Any]) -> dict[str, Any]:
+    return {"status": "ok", "event": bus.publish(**payload)}
+
+
+def list_agent_events(kind: str | None = None, status: str = "pending") -> dict[str, Any]:
+    return {"status": "ok", "events": bus.list_events(kind, status)}
+
+
+def get_agent_event(event_id: str) -> dict[str, Any]:
+    return {"status": "ok", "event": bus.get(event_id)}
+
+
+def transition_agent_event(event_id: str, status: str, actor: str = "system") -> dict[str, Any]:
+    return {"status": "ok", "event": bus.transition(event_id, status, actor)}
+
+
+def agent_event_stats() -> dict[str, Any]:
+    return {"status": "ok", "stats": bus.stats()}
+
+
+# Compatibility adapters
+
+def list_gpt_inbox_messages():
+    return list_agent_events(kind="gpt_inbox")
+
+
+def ack_gpt_inbox_message(event_id: str):
+    return transition_agent_event(event_id, "completed", "custom-gpt")
+
+
+def build_router():
+    if APIRouter is None:
+        return None
+    router = APIRouter(prefix="/agent-events", tags=["agent-events"])
+
+    @router.post("", operation_id="createAgentEvent")
+    def create(payload: dict[str, Any]):
+        return create_agent_event(payload)
+
+    @router.get("", operation_id="listAgentEvents")
+    def listing(kind: str | None = None, status: str = "pending"):
+        return list_agent_events(kind, status)
+
+    @router.get("/{event_id}", operation_id="getAgentEvent")
+    def get(event_id: str):
+        return get_agent_event(event_id)
+
+    @router.post("/{event_id}/transition", operation_id="transitionAgentEvent")
+    def transition(event_id: str, payload: dict[str, Any]):
+        return transition_agent_event(event_id, payload["status"], payload.get("actor", "system"))
+
+    @router.get("/stats/overview", operation_id="getAgentEventStats")
+    def stats():
+        return agent_event_stats()
+
+    return router
+
+
+router = build_router()
+
+__all__ = ["router", "create_agent_event", "list_agent_events", "get_agent_event", "transition_agent_event"]
@@ -0,0 +1,125 @@
+"""Unified Agent Event Bus.
+
+Unifies commands, GPT messages, replies, approvals, audit and reports into one
+state machine. Compatibility layers can filter events by kind instead of keeping
+separate queues.
+"""
+
+from __future__ import annotations
+
+from dataclasses import asdict, dataclass, field
+from datetime import datetime, timezone
+from hashlib import sha256
+from pathlib import Path
+import json
+from typing import Any
+
+
+def utc_now() -> str:
+    return datetime.now(timezone.utc).isoformat()
+
+
+@dataclass
+class AgentEvent:
+    id: str
+    kind: str
+    source: str
+    target: str
+    summary: str
+    payload: dict[str, Any]
+    status: str = "pending"
+    correlation_id: str | None = None
+    created_at: str = field(default_factory=utc_now)
+    updated_at: str = field(default_factory=utc_now)
+    actor: str | None = None
+
+    def to_dict(self) -> dict[str, Any]:
+        return asdict(self)
+
+
+class AgentEventBus:
+    STATES = ("pending", "running", "completed", "failed")
+
+    def __init__(self, root: str = "agents/events"):
+        self.root = Path(root)
+        for state in self.STATES:
+            (self.root / state).mkdir(parents=True, exist_ok=True)
+        (self.root / "audit").mkdir(parents=True, exist_ok=True)
+
+    def publish(
+        self,
+        kind: str,
+        source: str,
+        target: str,
+        summary: str,
+        payload: dict[str, Any],
+        correlation_id: str | None = None,
+    ) -> dict[str, Any]:
+        event_id = "evt_" + sha256(
+            json.dumps(
+                {"kind": kind, "source": source, "summary": summary, "payload": payload},
+                sort_keys=True,
+                ensure_ascii=False,
+            ).encode("utf-8")
+        ).hexdigest()[:16]
+        event = AgentEvent(event_id, kind, source, target, summary, payload, correlation_id=correlation_id)
+        self._write("pending", event.to_dict())
+        self._audit("created", event.to_dict())
+        return event.to_dict()
+
+    def list_events(self, kind: str | None = None, status: str = "pending", limit: int = 100) -> list[dict[str, Any]]:
+        result = []
+        for path in sorted((self.root / status).glob("*.json")):
+            event = self._read(path)
+            if kind and event.get("kind") != kind:
+                continue
+            result.append(event)
+        return result[:limit]
+
+    def get(self, event_id: str) -> dict[str, Any] | None:
+        for state in self.STATES:
+            path = self.root / state / f"{event_id}.json"
+            if path.exists():
+                return self._read(path)
+        return None
+
+    def transition(self, event_id: str, status: str, actor: str | None = None) -> dict[str, Any]:
+        event = self.get(event_id)
+        if not event:
+            return {"status": "not_found", "id": event_id}
+        if status not in self.STATES:
+            return {"status": "invalid_state"}
+        old = event.get("status", "pending")
+        old_path = self.root / old / f"{event_id}.json"
+        event["status"] = status
+        event["actor"] = actor
+        event["updated_at"] = utc_now()
+        self._write(status, event)
+        if old_path.exists() and old != status:
+            old_path.unlink()
+        self._audit("transition", event, {"from": old, "to": status})
+        return event
+
+    def stats(self) -> dict[str, Any]:
+        return {
+            "events": {state: len(list((self.root / state).glob("*.json"))) for state in self.STATES},
+            "audit_events": len(list((self.root / "audit").glob("*.json"))),
+            "status": "healthy",
+        }
+
+    def _write(self, state: str, data: dict[str, Any]):
+        (self.root / state / f"{data['id']}.json").write_text(
+            json.dumps(data, ensure_ascii=False, indent=2), encoding="utf-8"
+        )
+
+    def _read(self, path: Path):
+        return json.loads(path.read_text(encoding="utf-8"))
+
+    def _audit(self, action: str, event: dict[str, Any], extra: dict[str, Any] | None = None):
+        data = {"action": action, "event": event, "extra": extra or {}, "time": utc_now()}
+        (self.root / "audit" / f"{event['id']}-{action}.json").write_text(
+            json.dumps(data, ensure_ascii=False, indent=2), encoding="utf-8"
+        )
+
+
+__all__ = ["AgentEvent", "AgentEventBus"]
@@ -0,0 +1,50 @@
+"""Admin dashboard data model for the unified Agent Event Bus."""
+
+from __future__ import annotations
+
+from collections import Counter
+from typing import Any
+
+from .agent_event_bus import AgentEventBus
+
+
+class AgentEventDashboard:
+    def __init__(self, bus: AgentEventBus | None = None):
+        self.bus = bus or AgentEventBus()
+
+    def overview(self) -> dict[str, Any]:
+        stats = self.bus.stats()
+        pending = self.bus.list_events(status="pending")
+        running = self.bus.list_events(status="running")
+        completed = self.bus.list_events(status="completed")
+        failed = self.bus.list_events(status="failed")
+        all_visible = pending + running + completed + failed
+        by_kind = Counter(event.get("kind", "unknown") for event in all_visible)
+        by_target = Counter(event.get("target", "unknown") for event in all_visible)
+        return {
+            "module": "agent-event-bus",
+            "title": "Unified Agent Event Bus",
+            "route": "/admin/ai-evolution/events",
+            "metrics": {
+                **stats,
+                "by_kind": dict(by_kind),
+                "by_target": dict(by_target),
+                "gpt_inbox_pending": by_kind.get("gpt_inbox", 0),
+                "openclaw_pending": len([e for e in pending if e.get("target") == "openclaw"]),
+            },
+            "panels": [
+                "event timeline",
+                "pending commands",
+                "GPT inbox",
+                "agent replies",
+                "failure reports",
+                "audit trail",
+            ],
+            "audit": {
+                "enabled": True,
+                "location": "agents/events/audit",
+            },
+        }
+
+
+__all__ = ["AgentEventDashboard"]
@@ -0,0 +1,37 @@
+"""Bridge Agent Event Bus events into Agent Observability metrics."""
+
+from .agent_observability_hooks import AgentObservabilityHooks
+
+
+class AgentEventObservabilityBridge:
+    def __init__(self, hooks: AgentObservabilityHooks | None = None):
+        self.hooks = hooks or AgentObservabilityHooks()
+
+    def consume(self, event: dict):
+        kind = event.get("kind")
+        payload = event.get("payload", {})
+        if kind == "agent_started":
+            return self.hooks.start(
+                user_id=payload.get("user_id", "unknown"),
+                visible_scope=payload.get("scope", "self"),
+                entry_agent=payload.get("agent", "unknown"),
+                execution_id=event.get("id"),
+            )
+        if kind == "agent_delegated":
+            return self.hooks.delegate(event["execution_id"], payload.get("agent", "unknown"))
+        if kind == "token_consumed":
+            return self.hooks.tokens(
+                event["execution_id"],
+                payload.get("input_tokens", 0),
+                payload.get("output_tokens", 0),
+            )
+        if kind in {"agent_completed", "agent_failed"}:
+            return self.hooks.finish(
+                event["execution_id"],
+                payload.get("closed_by"),
+                "failed" if kind == "agent_failed" else "completed",
+            )
+        return {"status": "ignored", "kind": kind}
+
+
+__all__ = ["AgentEventObservabilityBridge"]
@@ -0,0 +1,80 @@
+"""Agent orchestration observability model.
+
+Tracks OpenClaw, Hermes and sub-agent execution chains. Token usage is optional
+and depends on model adapter instrumentation.
+"""
+
+from __future__ import annotations
+
+from dataclasses import asdict, dataclass, field
+from datetime import datetime, timezone
+from typing import Any
+
+
+def utc_now():
+    return datetime.now(timezone.utc).isoformat()
+
+
+@dataclass
+class AgentExecutionRecord:
+    execution_id: str
+    user_id: str
+    visible_scope: str
+    entry_agent: str
+    path: list[str]
+    used_hermes: bool
+    closed_by: str
+    duration_ms: int = 0
+    input_tokens: int | None = None
+    output_tokens: int | None = None
+    total_tokens: int | None = None
+    status: str = "completed"
+    started_at: str = field(default_factory=utc_now)
+    ended_at: str = field(default_factory=utc_now)
+
+    def to_dict(self):
+        return asdict(self)
+
+
+class AgentObservability:
+    def __init__(self):
+        self.records: list[AgentExecutionRecord] = []
+
+    def record(self, record: AgentExecutionRecord):
+        self.records.append(record)
+        return record.to_dict()
+
+    def summary(self, user_id: str | None = None) -> dict[str, Any]:
+        records = [r for r in self.records if not user_id or r.user_id == user_id]
+        hermes = [r for r in records if r.used_hermes]
+        direct = [r for r in records if not r.used_hermes]
+        return {
+            "total_executions": len(records),
+            "openclaw_independent_loops": len(direct),
+            "hermes_orchestrated_loops": len(hermes),
+            "average_duration_ms": (
+                sum(r.duration_ms for r in records) / len(records) if records else 0
+            ),
+            "token_usage": {
+                "input": sum(r.input_tokens or 0 for r in records),
+                "output": sum(r.output_tokens or 0 for r in records),
+                "total": sum(r.total_tokens or 0 for r in records),
+            },
+        }
+
+    def hermes_tree(self, user_id: str | None = None):
+        return [
+            {
+                "execution_id": r.execution_id,
+                "entry": r.entry_agent,
+                "used_hermes": r.used_hermes,
+                "children": r.path,
+                "duration_ms": r.duration_ms,
+                "tokens": r.total_tokens,
+            }
+            for r in self.records
+            if (not user_id or r.user_id == user_id)
+        ]
+
+
+__all__ = ["AgentExecutionRecord", "AgentObservability"]
@@ -0,0 +1,52 @@
+"""API for agent orchestration observability."""
+
+from __future__ import annotations
+
+from typing import Any
+
+try:
+    from fastapi import APIRouter
+except Exception:
+    APIRouter = None
+
+from .agent_observability import AgentExecutionRecord
+from .agent_observability_dashboard import AgentObservabilityDashboard
+from .agent_observability_store import AgentObservabilityStore
+
+store = AgentObservabilityStore()
+dashboard = AgentObservabilityDashboard(store.load())
+
+
+def record_agent_execution(payload: dict[str, Any]):
+    record = AgentExecutionRecord(**payload)
+    return {"status": "ok", "record": store.append(record)}
+
+
+def get_agent_observability(user_id: str | None = None, group_id: str | None = None, role: str = "user"):
+    records = store.query(user_id, group_id, role)
+    return {
+        "status": "ok",
+        "records": records,
+        "summary": store.load().summary(user_id if role != "owner" else None),
+    }
+
+
+def build_router():
+    if APIRouter is None:
+        return None
+    router = APIRouter(prefix="/admin/ai-evolution/agents", tags=["agent-observability"])
+
+    @router.post("/execution", operation_id="recordAgentExecution")
+    def record(payload: dict[str, Any]):
+        return record_agent_execution(payload)
+
+    @router.get("/overview", operation_id="getAgentObservabilityOverview")
+    def overview(user_id: str | None = None, group_id: str | None = None, role: str = "user"):
+        return get_agent_observability(user_id, group_id, role)
+
+    return router
+
+
+router = build_router()
+
+__all__ = ["router", "record_agent_execution", "get_agent_observability"]
@@ -0,0 +1,44 @@
+"""Web UI dashboard model for agent execution visibility."""
+
+from .agent_observability import AgentObservability
+
+
+class AgentObservabilityDashboard:
+    def __init__(self, observability: AgentObservability | None = None):
+        self.observability = observability or AgentObservability()
+
+    def overview(self, user_id: str | None = None):
+        return {
+            "route": "/admin/ai-evolution/agents",
+            "title": "Agent Orchestration Observability",
+            "summary": self.observability.summary(user_id),
+            "views": [
+                {
+                    "name": "scheduler_flow",
+                    "title": "OpenClaw/Hermes 调度流程",
+                    "data": self.observability.hermes_tree(user_id),
+                },
+                {
+                    "name": "agent_metrics",
+                    "title": "智能体指标",
+                    "fields": [
+                        "duration_ms",
+                        "input_tokens",
+                        "output_tokens",
+                        "total_tokens",
+                        "status",
+                    ],
+                },
+                {
+                    "name": "loop_distribution",
+                    "title": "闭环分布",
+                    "fields": [
+                        "openclaw_independent_loops",
+                        "hermes_orchestrated_loops",
+                    ],
+                },
+            ],
+        }
+
+
+__all__ = ["AgentObservabilityDashboard"]
@@ -0,0 +1,113 @@
+"""Runtime hooks for Agent Observability.
+
+OpenClaw/Hermes/sub-agents can call these hooks at start/end/delegation points.
+The hooks store sanitized metadata only: no prompts, no secrets, no private message
+content.
+"""
+
+from __future__ import annotations
+
+from dataclasses import dataclass, field
+from datetime import datetime, timezone
+from typing import Any
+from uuid import uuid4
+
+from .agent_observability import AgentExecutionRecord
+from .agent_observability_store import AgentObservabilityStore
+
+
+def utc_now() -> str:
+    return datetime.now(timezone.utc).isoformat()
+
+
+def parse_time(value: str) -> datetime:
+    return datetime.fromisoformat(value.replace("Z", "+00:00"))
+
+
+@dataclass
+class AgentRunContext:
+    execution_id: str
+    user_id: str
+    visible_scope: str
+    entry_agent: str
+    path: list[str] = field(default_factory=list)
+    used_hermes: bool = False
+    started_at: str = field(default_factory=utc_now)
+    input_tokens: int = 0
+    output_tokens: int = 0
+    status: str = "running"
+
+    def delegate(self, agent_name: str) -> None:
+        if agent_name not in self.path:
+            self.path.append(agent_name)
+        if agent_name.lower() == "hermes":
+            self.used_hermes = True
+
+    def add_tokens(self, input_tokens: int = 0, output_tokens: int = 0) -> None:
+        self.input_tokens += max(input_tokens, 0)
+        self.output_tokens += max(output_tokens, 0)
+
+
+class AgentObservabilityHooks:
+    def __init__(self, store: AgentObservabilityStore | None = None):
+        self.store = store or AgentObservabilityStore()
+        self.active: dict[str, AgentRunContext] = {}
+
+    def start(
+        self,
+        user_id: str,
+        visible_scope: str,
+        entry_agent: str,
+        execution_id: str | None = None,
+    ) -> dict[str, Any]:
+        run = AgentRunContext(
+            execution_id=execution_id or f"exec_{uuid4().hex[:12]}",
+            user_id=user_id,
+            visible_scope=visible_scope,
+            entry_agent=entry_agent,
+            path=[entry_agent],
+            used_hermes=entry_agent.lower() == "hermes",
+        )
+        self.active[run.execution_id] = run
+        return {"status": "started", "execution_id": run.execution_id, "run": run.__dict__}
+
+    def delegate(self, execution_id: str, agent_name: str) -> dict[str, Any]:
+        run = self.active[execution_id]
+        run.delegate(agent_name)
+        return {"status": "delegated", "execution_id": execution_id, "path": run.path}
+
+    def tokens(self, execution_id: str, input_tokens: int = 0, output_tokens: int = 0) -> dict[str, Any]:
+        run = self.active[execution_id]
+        run.add_tokens(input_tokens, output_tokens)
+        return {
+            "status": "tokens_recorded",
+            "execution_id": execution_id,
+            "input_tokens": run.input_tokens,
+            "output_tokens": run.output_tokens,
+            "total_tokens": run.input_tokens + run.output_tokens,
+        }
+
+    def finish(self, execution_id: str, closed_by: str | None = None, status: str = "completed") -> dict[str, Any]:
+        run = self.active.pop(execution_id)
+        ended_at = utc_now()
+        duration_ms = int((parse_time(ended_at) - parse_time(run.started_at)).total_seconds() * 1000)
+        record = AgentExecutionRecord(
+            execution_id=run.execution_id,
+            user_id=run.user_id,
+            visible_scope=run.visible_scope,
+            entry_agent=run.entry_agent,
+            path=run.path,
+            used_hermes=run.used_hermes,
+            closed_by=closed_by or run.path[-1],
+            duration_ms=duration_ms,
+            input_tokens=run.input_tokens,
+            output_tokens=run.output_tokens,
+            total_tokens=run.input_tokens + run.output_tokens,
+            status=status,
+            started_at=run.started_at,
+            ended_at=ended_at,
+        )
+        return {"status": "stored", "record": self.store.append(record)}
+
+
+__all__ = ["AgentRunContext", "AgentObservabilityHooks"]
@@ -0,0 +1,45 @@
+"""Persistent store for agent observability records.
+
+Stores sanitized execution metrics for OpenClaw/Hermes/sub-agents. No prompts,
+secrets, or private message contents are persisted here; only metrics and routing
+metadata are stored for Web UI statistics and audit.
+"""
+
+from __future__ import annotations
+
+import json
+from pathlib import Path
+from typing import Any
+
+from .agent_observability import AgentExecutionRecord, AgentObservability
+
+
+class AgentObservabilityStore:
+    def __init__(self, root: str = "runtime/agent-observability"):
+        self.root = Path(root)
+        self.records_dir = self.root / "records"
+        self.records_dir.mkdir(parents=True, exist_ok=True)
+
+    def append(self, record: AgentExecutionRecord) -> dict[str, Any]:
+        path = self.records_dir / f"{record.execution_id}.json"
+        data = record.to_dict()
+        path.write_text(json.dumps(data, ensure_ascii=False, indent=2), encoding="utf-8")
+        return data
+
+    def load(self) -> AgentObservability:
+        observability = AgentObservability()
+        for path in sorted(self.records_dir.glob("*.json")):
+            raw = json.loads(path.read_text(encoding="utf-8"))
+            observability.records.append(AgentExecutionRecord(**raw))
+        return observability
+
+    def query(self, user_id: str | None = None, group_id: str | None = None, role: str = "user") -> list[dict[str, Any]]:
+        records = [record.to_dict() for record in self.load().records]
+        if role == "owner":
+            return records
+        if role == "group_admin" and group_id:
+            return [r for r in records if r.get("visible_scope") == group_id or r.get("user_id") == user_id]
+        return [r for r in records if r.get("user_id") == user_id]
+
+
+__all__ = ["AgentObservabilityStore"]
@@ -15,7 +15,7 @@ TOKEN_FILE = Path(os.getenv('AIOS_GITOPS_TOKEN_FILE', str(SECRETS / 'gitops.toke
 HOST = os.getenv('AIOS_GITOPS_HOST', '127.0.0.1')
 PORT = int(os.getenv('AIOS_GITOPS_PORT', '18890'))
 OPENAPI = WS / 'docs' / 'gitops' / 'openapi-chatgpt-actions.yaml'
-BARE_REPO = Path(os.getenv('AIOS_GITOPS_BARE_REPO', str(ROOT / 'runtime' / 'gitea-bin' / 'data' / 'repositories' / 'aios' / 'aios-v23.git')))
+BARE_REPO = Path(os.getenv('AIOS_GITOPS_BARE_REPO', str(ROOT / 'runtime' / 'gitea-bin' / 'data' / 'repositories' / 'aios' / 'aios.git')))
 monitor = ActionMonitor()

 BASE_MODELS = [
@@ -200,6 +200,11 @@ class H(BaseHTTPRequestHandler):
                latest=monitor.latest()
                ldict={'event_id':latest.event_id,'action':latest.action,'error':latest.error,'context':latest.context,'created_at':latest.created_at} if latest else None
                return self.sendj({'events':[{"event_id":e.event_id,"action":e.action,"error":e.error,"context":e.context,"created_at":e.created_at} for e in monitor.list_events()],'latest':ldict})
+                        # Admin AI Evolution Dashboard
+            if path in ('/admin/ai-evolution','/admin/ai-evolution/'):
+                return self.ai_evolution_page()
+            if path in ('/admin/ai-evolution/api/data'):
+                return self.ai_evolution_data()
            return self.sendj({'error':'not_found','path':path},404)
        except Exception as e:
            monitor.record_failure('do_GET', str(e), {'path':path})
@@ -290,9 +295,102 @@ class H(BaseHTTPRequestHandler):
        elif 'model' in low or '模型' in text: intent='model_change'; plan=['use POST /api/model or PUT /api/model-priority','commit metadata']
        elif 'patch' in low or '补丁' in text: intent='patch_create'; plan=['call POST /api/patch']
        b=checkout(data.get('branch') or f'patch/nl-{uid("nl")}', 'main', True); doc={'id':uid('nl'),'text':text,'intent':intent,'action_plan':plan,'execute':False,'owner_approval_required':True,'created_by':a['account'],'created_at':now()}; r=write_json(f'patches/nl/{safe(doc["id"])}.json', doc); c=commit([r], f'chore(chatgpt): record nl plan {intent}'); self.sendj({'status':'planned','branch':b,'request':doc,'commit':c})
+    _EVOLUTION_HTML = '''<!DOCTYPE html>
+<html lang=zh-CN>
+<head>
+<meta charset=UTF-8>
+<meta name=viewport content="width=device-width,initial-scale=1.0">
+<title>AIOS Evolution Dashboard</title>
+<script src=https://cdn.jsdelivr.net/npm/vue@3/dist/vue.global.prod.js></script>
+<style>
+*{margin:0;padding:0;box-sizing:border-box}
+body{background:#0d1117;color:#c9d1d9;font-size:14px;line-height:1.7;font-family:-apple-system,BlinkMacSystemFont,Segoe UI,Helvetica,Arial,sans-serif}
+.c{max-width:1000px;margin:0 auto;padding:32px 24px}
+h1{font-size:24px;color:#58a6ff;margin-bottom:2px}
+.sub{color:#8b949e;font-size:13px;border-bottom:1px solid #21262d;padding-bottom:16px;margin-bottom:20px;display:flex;justify-content:space-between}
+.g{display:grid;grid-template-columns:repeat(auto-fill,minmax(160px,1fr));gap:16px;margin-bottom:20px}
+.s{background:#161b22;border:1px solid #21262d;border-radius:8px;padding:18px;text-align:center}
+.s .v{font-size:28px;font-weight:700;color:#f0f6fc}
+.s .l{font-size:11px;color:#8b949e;margin-top:4px}
+.bx{background:#161b22;border:1px solid #21262d;border-radius:8px;margin-bottom:14px}
+.bh{border-bottom:1px solid #21262d;padding:10px 16px;font-weight:600;font-size:14px;display:flex;gap:6px;align-items:center}
+.bd{padding:12px 16px}
+.rw{display:grid;grid-template-columns:1fr 1fr 100px 60px;gap:8px;padding:8px 16px;border-bottom:1px solid #21262d;font-size:13px;align-items:center;cursor:pointer;transition:background .15s}
+.rw:hover{background:#1c2128}.rw:last-child{border-bottom:none}
+.tg{display:inline-block;padding:2px 10px;border-radius:99px;font-size:11px;font-weight:600}
+.ok{background:#3fb95033;color:#3fb950}.fl{background:#da363333;color:#f85149}.if{background:#1f6feb33;color:#58a6ff}
+pre{background:#0d1117;padding:12px;border-radius:6px;overflow:auto;font-size:12px;color:#8b949e;max-height:300px;margin-top:8px}select{padding:8px 10px;border:1px solid #30363d;border-radius:6px;background:#21262d;color:#c9d1d9;font-size:13px;max-width:400px;width:100%;margin-bottom:4px;cursor:pointer}
+</style>
+</head>
+<body>
+<div id=app class=c>
+<h1>AIOS Evolution Dashboard</h1>
+<div class=sub><span>Baseline: {{baseline}}</span><span style=color:#8b949e>{{refreshMsg}}</span></div>
+<div class=g><div class=s><div class=v>{{tot}}</div><div class=l>Commands</div></div><div class=s><div class=v style=color:#3fb950>{{sok}}</div><div class=l>Succeeded</div></div><div class=s><div class=v style=color:#f85149>{{sfail}}</div><div class=l>Failed</div></div><div class=s><div class=v>{{runs}}</div><div class=l>Executions</div></div></div>
+<div class=bx><div class=bh>Version Timeline</div><div class=bd><div class=rw style=font-weight:600;border-bottom:2px solid #30363d;cursor:default><span>Version</span><span>Time</span><span>Result</span></div><div v-for=v in vs class=rw @click=selVer(v)><span style=color:#58a6ff>{{v.tag}}</span><span>{{v.time||'-'}}</span><span :class="'tg '+v.result">{{v.result||'-'}}</span></div></div></div>
+<div class=bx><div class=bh>Commands</div><div class=bd><select v-model=cmd @change=loadCmd><option value>-- Select --</option><option v-for=c in cmds :value=c>{{c}}</option></select><pre v-if=cdet>{{cdet}}</pre></div></div>
+<div class=bx><div class=bh>Recent Agent Runs</div><div class=bd><div class=rw style=font-weight:600;border-bottom:2px solid #30363d;cursor:default><span>Agent</span><span>Command</span><span>Count</span><span>Result</span></div><div v-for=r in aruns class=rw><span>{{r.agent||'-'}}</span><span style=overflow:hidden;white-space:nowrap;text-overflow:ellipsis>{{r.command}}</span><span>{{r.interaction_count||0}}</span><span :class="'tg '+r.final_result">{{r.final_result||'-'}}</span></div></div></div></div>
+<script>
+const{createApp,ref,onMounted}=Vue;createApp({setup(){const vs=ref([]),cmds=ref([]),ar=ref([]),cmd=ref(''),cdet=ref(null),tot=ref(0),sok=ref(0),sfail=ref(0),runs=ref(0),baseline=ref('V1R3C1'),refreshMsg=ref('')
+async function ld(){try{const r=await fetch('/admin/ai-evolution/api/data'),d=await r.json();baseline.value=d.baseline||'V1R3C1';vs.value=d.versions||[];cmds.value=d.commands||[];ar.value=d.agent_runs||[];tot.value=d.stats?.total||0;sok.value=d.stats?.success||0;sfail.value=d.stats?.failed||0;runs.value=d.stats?.agent_runs||0;refreshMsg.value='Last updated: '+new Date().toLocaleString()}catch(e){}}
+async function loadCmd(){if(!cmd.value){cdet.value=null;return}try{const r=await fetch('/admin/ai-evolution/api/data?command='+encodeURIComponent(cmd.value)),d=await r.json();cdet.value=JSON.stringify(d,null,2)}catch(e){cdet.value='Error'}}
+function selVer(v){cmd.value=v.tag||v.id||v.version;loadCmd()}
+onMounted(()=>{ld();setInterval(ld,30000)})
+return{vs,cmds,ar,cmd,cdet,tot,sok,sfail,runs,baseline,refreshMsg,loadCmd,selVer}})mount('#app')
+</script></body></html>'''
+
+    def ai_evolution_page(self):
+        self.send_response(200)
+        self.send_header('Content-Type', 'text/html; charset=utf-8')
+        self.send_header('Content-Length', str(len(self._EVOLUTION_HTML)))
+        self.end_headers()
+        self.wfile.write(self._EVOLUTION_HTML.encode())
+
+    def ai_evolution_data(self):
+        q=urllib.parse.urlparse(self.path).query
+        params=urllib.parse.parse_qs(q)
+        cmd_filter=(params.get('command') or [None])[0]
+        bl=jload(WS/'baseline'/'current.json',{})
+        current_ver=bl.get('current_baseline','V1R3C1')
+        tags=sh(['tag','--list','V*'])['stdout'].splitlines() if WS.exists() else []
+        versions=[]
+        if (WS/'baseline'/'promotions').exists():
+            for p in sorted((WS/'baseline'/'promotions').glob('*.json')):
+                d=jload(p,{});v=d.get('version',p.stem)
+                versions.append({'tag':v,'time':d.get('promoted_at','-'),'result':'ok'})
+        for t in tags:
+            if not any(v['tag']==t for v in versions):
+                versions.append({'tag':t,'time':'-','result':'if'})
+        agent_runs=[];ard=WS/'audit'/'agent-runs'
+        if ard.exists():
+            for f in sorted(ard.glob('*.json'),reverse=True)[:50]:
+                d=jload(f,{});agent_runs.append(d) if isinstance(d,dict) else None
+        cmd_dirs=[WS/'agents/commands/active',WS/'agents/commands/running',WS/'agents/commands/completed',WS/'agents/commands/failed']
+        commands=[]
+        for d in cmd_dirs:
+            if d.exists():
+                for f in sorted(d.glob('*.md')):
+                    commands.append(f.name)
+        command_detail=None
+        if cmd_filter:
+            for d in cmd_dirs:
+                if d.exists():
+                    fp=d/cmd_filter
+                    if fp.exists():
+                        command_detail={'command':cmd_filter,'content':fp.read_text(encoding='utf-8')}
+                        break
+            if not command_detail:
+                command_detail={'command':cmd_filter,'error':'not_found'}
+            return self.sendj(command_detail)
+        completed=list((WS/'agents/commands/completed').glob('*.md')) if (WS/'agents/commands/completed').exists() else []
+        failed=list((WS/'agents/commands/failed').glob('*.md')) if (WS/'agents/commands/failed').exists() else []
+        stats={'total':len(commands),'success':len(completed),'failed':len(failed),'agent_runs':len(agent_runs)}
+        data={'baseline':current_ver,'versions':versions[-30:],'agent_runs':agent_runs,'commands':sorted(set(commands)),'stats':stats}
+        self.sendj(data)
+

 def api_spec():
-    return {'version':'V1R1C0-chatgpt-actions','openapi':'/api/openapi.yaml','read':['GET /api/base','GET /api/ver','GET /api/diff','GET /api/patch','GET /api/model','GET /api/run'],'write':['POST /api/branch','POST /api/file','POST /api/commit','POST /api/patch','POST /api/tag','POST /api/rollback','POST /api/model','PUT /api/model-priority','POST /api/nl'],'policy':{'bearer_required_for_write':True,'direct_main_push':False,'delete_tag':False,'owner_approval_required_for_v_level':True,'api_keys_in_git':False,'git_secret_field':'secret_ref','online_git_sync':True}}
+    return {'version':'V1R1C0-chatgpt-actions','openapi':'/api/openapi.yaml','read':['GET /api/base','GET /api/ver','GET /api/diff','GET /api/patch','GET /api/model','GET /api/run','GET /admin/ai-evolution'],'write':['POST /api/branch','POST /api/file','POST /api/commit','POST /api/patch','POST /api/tag','POST /api/rollback','POST /api/model','PUT /api/model-priority','POST /api/nl'],'policy':{'bearer_required_for_write':True,'direct_main_push':False,'delete_tag':False,'owner_approval_required_for_v_level':True,'api_keys_in_git':False,'git_secret_field':'secret_ref','online_git_sync':True}}

 def main():
    print(f'AIOS ChatGPT GitOps adapter listening on {HOST}:{PORT}')
@@ -0,0 +1,120 @@
+"""Architect governance agent responsibilities.
+
+This agent records requirement changes, maintains versioned human-readable
+requirement/architecture/deployment/test documents, reviews architecture changes,
+validates deployment docs against actual configuration, and produces version bump
+recommendations.
+"""
+
+from __future__ import annotations
+
+from dataclasses import asdict, dataclass, field
+from datetime import datetime, timezone
+from typing import Any
+
+
+def utc_now() -> str:
+    return datetime.now(timezone.utc).isoformat()
+
+
+@dataclass(frozen=True)
+class RequirementChange:
+    id: str
+    source: str
+    actor: str
+    requirement: str
+    created_at: str = field(default_factory=utc_now)
+
+    def to_dict(self) -> dict[str, Any]:
+        return asdict(self)
+
+
+@dataclass(frozen=True)
+class DocumentRefreshPlan:
+    requirement_doc: bool
+    architecture_doc: bool
+    deployment_doc: bool
+    test_case_doc: bool
+    reason: str
+
+    def to_dict(self) -> dict[str, Any]:
+        return asdict(self)
+
+
+@dataclass(frozen=True)
+class ArchitectureReviewDecision:
+    status: str
+    findings: list[str]
+    optimization_recommendations: list[str]
+    version_bump: str
+    docs_to_refresh: DocumentRefreshPlan
+    created_at: str = field(default_factory=utc_now)
+
+    def to_dict(self) -> dict[str, Any]:
+        return {
+            "status": self.status,
+            "findings": self.findings,
+            "optimization_recommendations": self.optimization_recommendations,
+            "version_bump": self.version_bump,
+            "docs_to_refresh": self.docs_to_refresh.to_dict(),
+            "created_at": self.created_at,
+        }
+
+
+class ArchitectGovernanceAgent:
+    """Independent architecture governance role."""
+
+    def record_requirement_change(self, source: str, actor: str, requirement: str) -> RequirementChange:
+        change_id = "REQ-" + datetime.now(timezone.utc).strftime("%Y%m%d%H%M%S")
+        return RequirementChange(change_id, source, actor, requirement)
+
+    def review_architecture_change(self, proposal: dict[str, Any]) -> ArchitectureReviewDecision:
+        impact = proposal.get("impact", "medium")
+        findings = [
+            "requirement_traceability_checked",
+            "architecture_consistency_checked",
+            "deployment_doc_alignment_checked",
+            "test_case_doc_required",
+            "rollback_and_owner_gate_checked",
+        ]
+        recommendations = [
+            "refresh user requirements document",
+            "refresh architecture design document",
+            "refresh engineering deployment document",
+            "generate acceptance test case document",
+        ]
+        version_bump = self.decide_version_bump(proposal)
+        return ArchitectureReviewDecision(
+            status="review_required" if impact in {"high", "architecture"} else "pass",
+            findings=findings,
+            optimization_recommendations=recommendations,
+            version_bump=version_bump,
+            docs_to_refresh=DocumentRefreshPlan(True, True, True, True, "architecture or requirement changed"),
+        )
+
+    def decide_version_bump(self, proposal: dict[str, Any]) -> str:
+        if proposal.get("breaks_architecture") or proposal.get("major_capability"):
+            return "V"
+        if proposal.get("new_release_capability") or proposal.get("architecture_change"):
+            return "R"
+        return "C"
+
+    def validate_deployment_doc(self, deployment_doc: dict[str, Any], actual_config: dict[str, Any]) -> dict[str, Any]:
+        expected = set(deployment_doc.get("services", []))
+        actual = set(actual_config.get("services", []))
+        missing = sorted(expected - actual)
+        extra = sorted(actual - expected)
+        return {
+            "status": "pass" if not missing else "attention_required",
+            "missing_services": missing,
+            "extra_services": extra,
+            "checked_at": utc_now(),
+        }
+
+
+__all__ = [
+    "RequirementChange",
+    "DocumentRefreshPlan",
+    "ArchitectureReviewDecision",
+    "ArchitectGovernanceAgent",
+]
@@ -0,0 +1,32 @@
+"""API for Architect Governance Center."""
+
+try:
+    from fastapi import APIRouter
+except Exception:
+    APIRouter = None
+
+from .architect_governance_dashboard import ArchitectGovernanceDashboard
+
+
+dashboard = ArchitectGovernanceDashboard()
+
+
+def get_architect_governance_overview():
+    return dashboard.overview()
+
+
+def build_router():
+    if APIRouter is None:
+        return None
+    router = APIRouter(prefix="/admin/ai-evolution/architect", tags=["architect-governance"])
+
+    @router.get("/overview", operation_id="getArchitectGovernanceOverview")
+    def overview():
+        return get_architect_governance_overview()
+
+    return router
+
+
+router = build_router()
+
+__all__ = ["router", "get_architect_governance_overview"]
@@ -0,0 +1,153 @@
+"""Web UI dashboard model for Architect Governance Agent.
+
+Shows requirement changes, architecture review findings, document refresh status,
+deployment consistency, test case readiness, weekly research and V/R/C version
+recommendations.
+"""
+
+from __future__ import annotations
+
+from dataclasses import asdict, dataclass, field
+from datetime import datetime, timezone
+from typing import Any
+
+
+def utc_now() -> str:
+    return datetime.now(timezone.utc).isoformat()
+
+
+@dataclass(frozen=True)
+class ArchitectPanelMetric:
+    key: str
+    title: str
+    value: Any
+    status: str = "ok"
+    detail: str = ""
+
+    def to_dict(self) -> dict[str, Any]:
+        return asdict(self)
+
+
+@dataclass(frozen=True)
+class ArchitectPanel:
+    key: str
+    title: str
+    description: str
+    metrics: list[ArchitectPanelMetric] = field(default_factory=list)
+
+    def to_dict(self) -> dict[str, Any]:
+        return {
+            "key": self.key,
+            "title": self.title,
+            "description": self.description,
+            "metrics": [m.to_dict() for m in self.metrics],
+        }
+
+
+class ArchitectGovernanceDashboard:
+    def overview(
+        self,
+        requirement_changes: list[dict[str, Any]] | None = None,
+        architecture_reviews: list[dict[str, Any]] | None = None,
+        document_status: dict[str, Any] | None = None,
+        research_status: dict[str, Any] | None = None,
+        version_recommendations: list[dict[str, Any]] | None = None,
+    ) -> dict[str, Any]:
+        requirement_changes = requirement_changes or []
+        architecture_reviews = architecture_reviews or []
+        document_status = document_status or {}
+        research_status = research_status or {}
+        version_recommendations = version_recommendations or []
+        panels = [
+            self.requirement_panel(requirement_changes),
+            self.architecture_panel(architecture_reviews),
+            self.document_panel(document_status),
+            self.research_panel(research_status),
+            self.version_panel(version_recommendations),
+        ]
+        return {
+            "module": "architect-governance",
+            "route": "/admin/ai-evolution/architect",
+            "title": "架构师 Agent 治理中心",
+            "status": "attention_required" if any(self._panel_warn(panel) for panel in panels) else "healthy",
+            "generated_at": utc_now(),
+            "panels": [panel.to_dict() for panel in panels],
+            "requirement_changes": requirement_changes,
+            "architecture_reviews": architecture_reviews,
+            "version_recommendations": version_recommendations,
+        }
+
+    def requirement_panel(self, changes: list[dict[str, Any]]) -> ArchitectPanel:
+        return ArchitectPanel(
+            key="requirements",
+            title="需求变化记录",
+            description="记录来自 Custom GPT、Codex、OpenClaw、Hermes 和用户反馈的需求变化。",
+            metrics=[
+                ArchitectPanelMetric("total_requirement_changes", "需求变化数量", len(changes)),
+                ArchitectPanelMetric(
+                    "undocumented_changes",
+                    "待刷新到用户需求文档",
+                    len([c for c in changes if not c.get("documented")]),
+                    "warn" if any(not c.get("documented") for c in changes) else "ok",
+                ),
+            ],
+        )
+
+    def architecture_panel(self, reviews: list[dict[str, Any]]) -> ArchitectPanel:
+        pending = [r for r in reviews if r.get("status") not in {"pass", "not_required"}]
+        return ArchitectPanel(
+            key="architecture_review",
+            title="全量架构审视",
+            description="展示架构一致性、扩展性、安全边界、可运维性和回滚策略审查。",
+            metrics=[
+                ArchitectPanelMetric("review_count", "架构审查数量", len(reviews)),
+                ArchitectPanelMetric("pending_reviews", "待处理架构审查", len(pending), "warn" if pending else "ok"),
+            ],
+        )
+
+    def document_panel(self, status: dict[str, Any]) -> ArchitectPanel:
+        required_docs = ["requirements", "architecture", "deployment", "test_cases"]
+        missing = [doc for doc in required_docs if status.get(doc) not in {"ready", "updated"}]
+        return ArchitectPanel(
+            key="documents",
+            title="文档版本化状态",
+            description="用户需求、架构设计、工程部署、测试例文档刷新状态。",
+            metrics=[
+                ArchitectPanelMetric("requirements_doc", "用户需求文档", status.get("requirements", "unknown"), "warn" if "requirements" in missing else "ok"),
+                ArchitectPanelMetric("architecture_doc", "架构设计文档", status.get("architecture", "unknown"), "warn" if "architecture" in missing else "ok"),
+                ArchitectPanelMetric("deployment_doc", "工程部署文档", status.get("deployment", "unknown"), "warn" if "deployment" in missing else "ok"),
+                ArchitectPanelMetric("test_case_doc", "测试例文档", status.get("test_cases", "unknown"), "warn" if "test_cases" in missing else "ok"),
+            ],
+        )
+
+    def research_panel(self, status: dict[str, Any]) -> ArchitectPanel:
+        overdue = bool(status.get("overdue"))
+        return ArchitectPanel(
+            key="weekly_research",
+            title="每周外部技术情报",
+            description="联网搜索 Agent/模型版本、GitHub top 实现、潜在安全影响。",
+            metrics=[
+                ArchitectPanelMetric("last_research_at", "最近调研时间", status.get("last_research_at", "never"), "warn" if overdue else "ok"),
+                ArchitectPanelMetric("findings", "发现数量", status.get("findings", 0)),
+                ArchitectPanelMetric("optimization_candidates", "可借鉴优化项", status.get("optimization_candidates", 0)),
+            ],
+        )
+
+    def version_panel(self, recommendations: list[dict[str, Any]]) -> ArchitectPanel:
+        latest = recommendations[-1] if recommendations else {}
+        return ArchitectPanel(
+            key="version_decision",
+            title="V/R/C 版本建议",
+            description="由架构师 Agent 判断变更应进入 C、R 或 V。",
+            metrics=[
+                ArchitectPanelMetric("recommendation_count", "版本建议数量", len(recommendations)),
+                ArchitectPanelMetric("latest_recommendation", "最近建议", latest.get("version_bump", "none")),
+                ArchitectPanelMetric("reason", "建议原因", latest.get("reason", "")),
+            ],
+        )
+
+    def _panel_warn(self, panel: ArchitectPanel) -> bool:
+        return any(metric.status == "warn" for metric in panel.metrics)
+
+
+__all__ = ["ArchitectPanelMetric", "ArchitectPanel", "ArchitectGovernanceDashboard"]
@@ -0,0 +1,40 @@
+"""Web UI model for Architect weekly research."""
+
+from .architect_weekly_research_agent import ArchitectWeeklyResearchAgent
+
+
+class ArchitectResearchDashboard:
+    def __init__(self, agent=None):
+        self.agent = agent or ArchitectWeeklyResearchAgent()
+
+    def overview(self):
+        return {
+            "module": "architect-weekly-research",
+            "route": "/admin/ai-evolution/architect/research",
+            "title": "架构师 Agent 技术情报中心",
+            "panels": [
+                {
+                    "title": "Research Targets",
+                    "data": self.agent.plan(),
+                },
+                {
+                    "title": "Evaluation",
+                    "fields": [
+                        "relevance",
+                        "impact",
+                        "recommendation",
+                    ],
+                },
+                {
+                    "title": "Version Decision",
+                    "fields": [
+                        "C",
+                        "R",
+                        "V",
+                    ],
+                },
+            ],
+        }
+
+
+__all__ = ["ArchitectResearchDashboard"]
@@ -0,0 +1,122 @@
+"""Architect research timeline dashboard.
+
+Shows weekly external research findings as a traceable architecture-impact flow:
+research target -> finding -> impact -> recommendation -> version decision ->
+governance event.
+"""
+
+from __future__ import annotations
+
+from dataclasses import asdict, dataclass, field
+from datetime import datetime, timezone
+from typing import Any
+
+
+def utc_now() -> str:
+    return datetime.now(timezone.utc).isoformat()
+
+
+@dataclass(frozen=True)
+class ResearchTimelineStep:
+    id: str
+    title: str
+    step_type: str
+    status: str
+    source: str
+    timestamp: str
+    impact: str = "none"
+    version_bump: str = "none"
+    detail: dict[str, Any] = field(default_factory=dict)
+
+    def to_dict(self) -> dict[str, Any]:
+        return asdict(self)
+
+
+class ArchitectResearchTimeline:
+    ORDER = {
+        "research_planned": 10,
+        "finding_created": 20,
+        "impact_analyzed": 30,
+        "recommendation_created": 40,
+        "version_decision_created": 50,
+        "architecture_review_event_created": 60,
+    }
+
+    def build(self, findings: list[dict[str, Any]] | None = None, decisions: list[dict[str, Any]] | None = None) -> dict[str, Any]:
+        findings = findings or []
+        decisions = decisions or []
+        steps: list[ResearchTimelineStep] = [
+            ResearchTimelineStep(
+                id="research-plan",
+                title="每周技术情报调研计划",
+                step_type="research_planned",
+                status="ok",
+                source="architect-agent",
+                timestamp=utc_now(),
+                detail={"cadence": "weekly"},
+            )
+        ]
+        for idx, finding in enumerate(findings):
+            steps.append(
+                ResearchTimelineStep(
+                    id=finding.get("id", f"finding-{idx}"),
+                    title=finding.get("title", "research finding"),
+                    step_type="finding_created",
+                    status="ok",
+                    source=finding.get("source", "unknown"),
+                    timestamp=finding.get("created_at", utc_now()),
+                    impact=finding.get("impact", "none"),
+                    detail=finding,
+                )
+            )
+            steps.append(
+                ResearchTimelineStep(
+                    id=f"impact-{finding.get('id', idx)}",
+                    title="架构影响分析",
+                    step_type="impact_analyzed",
+                    status="attention_required" if finding.get("impact") in {"architecture", "breaking", "new_capability"} else "ok",
+                    source="architect-agent",
+                    timestamp=utc_now(),
+                    impact=finding.get("impact", "none"),
+                    detail={"relevance": finding.get("relevance"), "recommendation": finding.get("recommendation")},
+                )
+            )
+        for idx, decision in enumerate(decisions):
+            steps.append(
+                ResearchTimelineStep(
+                    id=decision.get("id", f"decision-{idx}"),
+                    title="C/R/V 版本建议",
+                    step_type="version_decision_created",
+                    status=decision.get("status", "ok"),
+                    source="architect-agent",
+                    timestamp=decision.get("created_at", utc_now()),
+                    version_bump=decision.get("version_bump", "none"),
+                    detail=decision,
+                )
+            )
+        steps.sort(key=lambda s: (self.ORDER.get(s.step_type, 999), s.timestamp))
+        return {
+            "module": "architect-research-timeline",
+            "route": "/admin/ai-evolution/architect/research/timeline",
+            "title": "架构师技术情报时间线",
+            "status": self._status(steps),
+            "generated_at": utc_now(),
+            "steps": [step.to_dict() for step in steps],
+            "summary": self._summary(steps),
+        }
+
+    def _status(self, steps: list[ResearchTimelineStep]) -> str:
+        if any(step.status == "attention_required" for step in steps):
+            return "attention_required"
+        return "healthy"
+
+    def _summary(self, steps: list[ResearchTimelineStep]) -> dict[str, Any]:
+        return {
+            "findings": len([s for s in steps if s.step_type == "finding_created"]),
+            "architecture_impacts": len([s for s in steps if s.impact in {"architecture", "breaking", "new_capability"}]),
+            "version_decisions": len([s for s in steps if s.step_type == "version_decision_created"]),
+            "recommended_bumps": sorted({s.version_bump for s in steps if s.version_bump != "none"}),
+        }
+
+
+__all__ = ["ResearchTimelineStep", "ArchitectResearchTimeline"]
@@ -0,0 +1,32 @@
+"""API for architect research timeline."""
+
+try:
+    from fastapi import APIRouter
+except Exception:
+    APIRouter = None
+
+from .architect_research_timeline import ArchitectResearchTimeline
+
+
+timeline = ArchitectResearchTimeline()
+
+
+def get_architect_research_timeline():
+    return timeline.build()
+
+
+def build_router():
+    if APIRouter is None:
+        return None
+    router = APIRouter(prefix="/admin/ai-evolution/architect/research", tags=["architect-research"])
+
+    @router.get("/timeline", operation_id="getArchitectResearchTimeline")
+    def get_timeline():
+        return get_architect_research_timeline()
+
+    return router
+
+
+router = build_router()
+
+__all__ = ["router", "get_architect_research_timeline"]
@@ -0,0 +1,113 @@
+"""Weekly external research loop for Architect Governance Agent.
+
+The agent plans and normalizes weekly research tasks. Actual internet access is
+provided by the runtime/search worker; this module defines the audited workflow,
+findings, impact analysis, and V/R/C recommendations.
+"""
+
+from __future__ import annotations
+
+from dataclasses import asdict, dataclass, field
+from datetime import datetime, timezone
+from typing import Any
+from uuid import uuid4
+
+
+def utc_now() -> str:
+    return datetime.now(timezone.utc).isoformat()
+
+
+@dataclass(frozen=True)
+class ResearchTarget:
+    id: str
+    topic: str
+    source_type: str
+    query: str
+    cadence: str = "weekly"
+
+    def to_dict(self) -> dict[str, Any]:
+        return asdict(self)
+
+
+@dataclass(frozen=True)
+class ResearchFinding:
+    id: str
+    target_id: str
+    title: str
+    summary: str
+    source: str
+    relevance: str
+    impact: str
+    recommendation: str
+    created_at: str = field(default_factory=utc_now)
+
+    def to_dict(self) -> dict[str, Any]:
+        return asdict(self)
+
+
+@dataclass(frozen=True)
+class ResearchDecision:
+    status: str
+    version_bump: str
+    reason: str
+    findings: list[ResearchFinding]
+    created_at: str = field(default_factory=utc_now)
+
+    def to_dict(self) -> dict[str, Any]:
+        return {
+            "status": self.status,
+            "version_bump": self.version_bump,
+            "reason": self.reason,
+            "findings": [f.to_dict() for f in self.findings],
+            "created_at": self.created_at,
+        }
+
+
+class ArchitectWeeklyResearchAgent:
+    DEFAULT_TARGETS = (
+        ResearchTarget("agent-frameworks", "Agent framework versions", "web", "latest multi-agent orchestration framework releases"),
+        ResearchTarget("model-ecosystem", "Model ecosystem changes", "web", "latest LLM model releases tool calling agent features"),
+        ResearchTarget("github-top-agentops", "GitHub top AgentOps implementations", "github", "top repositories agent observability multi agent orchestration"),
+        ResearchTarget("security-agent-risks", "Agent security changes", "web", "latest AI agent security sandbox prompt injection tool use risks"),
+    )
+
+    def plan(self) -> dict[str, Any]:
+        return {
+            "status": "planned",
+            "cadence": "weekly",
+            "targets": [target.to_dict() for target in self.DEFAULT_TARGETS],
+            "created_at": utc_now(),
+        }
+
+    def normalize_finding(
+        self,
+        target_id: str,
+        title: str,
+        summary: str,
+        source: str,
+        relevance: str = "medium",
+        impact: str = "none",
+        recommendation: str = "monitor",
+    ) -> ResearchFinding:
+        return ResearchFinding(
+            id=f"finding_{uuid4().hex[:12]}",
+            target_id=target_id,
+            title=title,
+            summary=summary,
+            source=source,
+            relevance=relevance,
+            impact=impact,
+            recommendation=recommendation,
+        )
+
+    def decide(self, findings: list[ResearchFinding]) -> ResearchDecision:
+        if any(f.impact == "breaking" for f in findings):
+            return ResearchDecision("action_required", "V", "breaking external ecosystem change", findings)
+        if any(f.impact in {"architecture", "new_capability"} for f in findings):
+            return ResearchDecision("optimization_recommended", "R", "external capability may improve architecture", findings)
+        if any(f.impact in {"bugfix", "minor"} for f in findings):
+            return ResearchDecision("minor_change_recommended", "C", "minor compatible improvement", findings)
+        return ResearchDecision("no_action", "none", "no relevant external change", findings)
+
+
+__all__ = ["ResearchTarget", "ResearchFinding", "ResearchDecision", "ArchitectWeeklyResearchAgent"]
@@ -0,0 +1,32 @@
+"""API for architecture change governance dashboard."""
+
+try:
+    from fastapi import APIRouter
+except Exception:
+    APIRouter = None
+
+from .architecture_change_dashboard import ArchitectureChangeDashboard
+
+
+dashboard = ArchitectureChangeDashboard()
+
+
+def get_architecture_change_overview():
+    return dashboard.overview()
+
+
+def build_router():
+    if APIRouter is None:
+        return None
+    router = APIRouter(prefix="/admin/ai-evolution/architecture-changes", tags=["architecture-changes"])
+
+    @router.get("/overview", operation_id="getArchitectureChangeOverview")
+    def overview():
+        return get_architecture_change_overview()
+
+    return router
+
+
+router = build_router()
+
+__all__ = ["router", "get_architecture_change_overview"]
@@ -0,0 +1,90 @@
+"""Architecture change classification.
+
+Small compatible architecture changes may be handled automatically by the
+Architect Agent with audit evidence. Large, risky, production-impacting or
+breaking architecture changes require explicit Owner approval before
+implementation.
+"""
+
+from __future__ import annotations
+
+from dataclasses import asdict, dataclass, field
+from typing import Any
+
+
+@dataclass(frozen=True)
+class ArchitectureChangeClassification:
+    level: str
+    owner_required: bool
+    auto_implement_allowed: bool
+    version_bump: str
+    reasons: list[str] = field(default_factory=list)
+    required_gates: list[str] = field(default_factory=list)
+
+    def to_dict(self) -> dict[str, Any]:
+        return asdict(self)
+
+
+class ArchitectureChangeClassifier:
+    """Classifies architecture changes into small/medium/large governance paths."""
+
+    LARGE_FLAGS = {
+        "breaking_change",
+        "breaks_architecture",
+        "production_architecture_update",
+        "security_boundary_change",
+        "cross_tenant_effect",
+        "cross_user_data_effect",
+        "data_migration",
+        "permission_model_change",
+        "runtime_infrastructure_change",
+    }
+
+    def classify(self, change: dict[str, Any]) -> ArchitectureChangeClassification:
+        risk = str(change.get("risk", "medium")).lower()
+        version_bump = str(change.get("version_bump", change.get("recommended_version_bump", "C"))).upper()
+        scope = str(change.get("scope", "component")).lower()
+        reasons: list[str] = []
+
+        for flag in sorted(self.LARGE_FLAGS):
+            if change.get(flag):
+                reasons.append(flag)
+
+        if risk in {"high", "critical"}:
+            reasons.append(f"risk={risk}")
+        if version_bump == "V":
+            reasons.append("version_bump=V")
+        if version_bump == "R" and scope in {"system", "platform", "production", "cross_agent", "cross_tenant"}:
+            reasons.append(f"version_bump=R scope={scope}")
+
+        if reasons:
+            return ArchitectureChangeClassification(
+                level="large",
+                owner_required=True,
+                auto_implement_allowed=False,
+                version_bump=version_bump,
+                reasons=reasons,
+                required_gates=["architect_review", "owner_architecture_approval", "test_gate", "promotion_gate"],
+            )
+
+        if risk == "low" and version_bump in {"C", "NONE"} and scope in {"component", "local", "documentation", "test"}:
+            return ArchitectureChangeClassification(
+                level="small",
+                owner_required=False,
+                auto_implement_allowed=True,
+                version_bump=version_bump,
+                reasons=["low_risk_compatible_change"],
+                required_gates=["architect_review", "audit", "test_gate"],
+            )
+
+        return ArchitectureChangeClassification(
+            level="medium",
+            owner_required=False,
+            auto_implement_allowed=False,
+            version_bump=version_bump,
+            reasons=["requires_architect_review_before_implementation"],
+            required_gates=["architect_review", "test_gate", "audit"],
+        )
+
+
+__all__ = ["ArchitectureChangeClassification", "ArchitectureChangeClassifier"]
@@ -0,0 +1,93 @@
+"""Architecture change governance dashboard.
+
+Displays architecture change classification, owner approval requirements,
+implementation readiness, version recommendation and audit evidence for Web UI.
+"""
+
+from __future__ import annotations
+
+from dataclasses import asdict, dataclass, field
+from datetime import datetime, timezone
+from typing import Any
+
+from .architecture_change_classifier import ArchitectureChangeClassifier
+
+
+def utc_now() -> str:
+    return datetime.now(timezone.utc).isoformat()
+
+
+@dataclass(frozen=True)
+class ArchitectureChangeView:
+    change_id: str
+    title: str
+    level: str
+    owner_required: bool
+    auto_implement_allowed: bool
+    version_bump: str
+    status: str
+    reasons: list[str] = field(default_factory=list)
+    required_gates: list[str] = field(default_factory=list)
+    owner_decision: str | None = None
+
+    def to_dict(self) -> dict[str, Any]:
+        return asdict(self)
+
+
+class ArchitectureChangeDashboard:
+    def __init__(self, classifier: ArchitectureChangeClassifier | None = None):
+        self.classifier = classifier or ArchitectureChangeClassifier()
+
+    def overview(self, changes: list[dict[str, Any]] | None = None) -> dict[str, Any]:
+        changes = changes or []
+        views = [self._view(change) for change in changes]
+        return {
+            "module": "architecture-change-governance",
+            "route": "/admin/ai-evolution/architecture-changes",
+            "title": "架构变更治理中心",
+            "status": self._status(views),
+            "generated_at": utc_now(),
+            "summary": self._summary(views),
+            "changes": [view.to_dict() for view in views],
+        }
+
+    def _view(self, change: dict[str, Any]) -> ArchitectureChangeView:
+        classification = self.classifier.classify(change)
+        owner_decision = change.get("owner_decision")
+        if classification.owner_required and owner_decision != "approved":
+            status = "waiting_owner_approval"
+        elif classification.auto_implement_allowed:
+            status = "auto_implementation_allowed"
+        else:
+            status = "ready_after_required_gates"
+        return ArchitectureChangeView(
+            change_id=change.get("change_id", change.get("id", "unknown")),
+            title=change.get("title", "architecture change"),
+            level=classification.level,
+            owner_required=classification.owner_required,
+            auto_implement_allowed=classification.auto_implement_allowed,
+            version_bump=classification.version_bump,
+            status=status,
+            reasons=classification.reasons,
+            required_gates=classification.required_gates,
+            owner_decision=owner_decision,
+        )
+
+    def _status(self, views: list[ArchitectureChangeView]) -> str:
+        if any(view.status == "waiting_owner_approval" for view in views):
+            return "attention_required"
+        return "healthy"
+
+    def _summary(self, views: list[ArchitectureChangeView]) -> dict[str, Any]:
+        return {
+            "total": len(views),
+            "small": len([v for v in views if v.level == "small"]),
+            "medium": len([v for v in views if v.level == "medium"]),
+            "large": len([v for v in views if v.level == "large"]),
+            "waiting_owner_approval": len([v for v in views if v.status == "waiting_owner_approval"]),
+            "auto_allowed": len([v for v in views if v.auto_implement_allowed]),
+            "version_bumps": sorted({v.version_bump for v in views if v.version_bump and v.version_bump != "NONE"}),
+        }
+
+
+__all__ = ["ArchitectureChangeView", "ArchitectureChangeDashboard"]
@@ -0,0 +1,89 @@
+"""Owner approval gate for architecture changes.
+
+Architecture changes may be proposed, reviewed and optimized by agents, but they
+must not be implemented until owner approval is recorded. This gate sits before
+implementation and before promotion.
+"""
+
+from __future__ import annotations
+
+from dataclasses import asdict, dataclass, field
+from datetime import datetime, timezone
+from typing import Any
+
+
+def utc_now() -> str:
+    return datetime.now(timezone.utc).isoformat()
+
+
+@dataclass(frozen=True)
+class ArchitectureOwnerApprovalRequest:
+    request_id: str
+    change_id: str
+    title: str
+    reason: str
+    risk: str
+    architect_decision: str
+    status: str = "pending_owner_approval"
+    created_at: str = field(default_factory=utc_now)
+
+    def to_dict(self) -> dict[str, Any]:
+        return asdict(self)
+
+
+@dataclass(frozen=True)
+class ArchitectureOwnerDecision:
+    request_id: str
+    change_id: str
+    decision: str
+    decided_by: str
+    reason: str = ""
+    decided_at: str = field(default_factory=utc_now)
+
+    def to_dict(self) -> dict[str, Any]:
+        return asdict(self)
+
+
+class ArchitectureOwnerGate:
+    """Blocks architecture implementation until owner explicitly approves."""
+
+    def create_request(self, change: dict[str, Any], architect_decision: dict[str, Any]) -> ArchitectureOwnerApprovalRequest:
+        return ArchitectureOwnerApprovalRequest(
+            request_id=f"ARCH-OWNER-{change.get('change_id', change.get('id', 'unknown'))}",
+            change_id=change.get("change_id", change.get("id", "unknown")),
+            title=change.get("title", "architecture change"),
+            reason=change.get("reason", "architecture impact requires owner approval"),
+            risk=change.get("risk", "medium"),
+            architect_decision=architect_decision.get("status", "reviewed"),
+        )
+
+    def can_implement(self, request: dict[str, Any], decision: dict[str, Any] | None = None) -> dict[str, Any]:
+        if not decision:
+            return {
+                "allowed": False,
+                "status": "waiting_owner_approval",
+                "reason": "architecture change requires owner approval before implementation",
+                "request": request,
+            }
+        if decision.get("decision") == "approved":
+            return {
+                "allowed": True,
+                "status": "owner_approved",
+                "reason": decision.get("reason", "approved by owner"),
+                "request": request,
+                "decision": decision,
+            }
+        return {
+            "allowed": False,
+            "status": "owner_rejected",
+            "reason": decision.get("reason", "rejected by owner"),
+            "request": request,
+            "decision": decision,
+        }
+
+
+__all__ = [
+    "ArchitectureOwnerApprovalRequest",
+    "ArchitectureOwnerDecision",
+    "ArchitectureOwnerGate",
+]
@@ -0,0 +1,79 @@
+"""HTTP handler and display data for architecture owner gate."""
+
+from __future__ import annotations
+
+from dataclasses import asdict
+from datetime import datetime, timezone
+from typing import Any
+
+from .architecture_owner_gate import ArchitectureOwnerGate, ArchitectureOwnerApprovalRequest
+
+
+gate = ArchitectureOwnerGate()
+
+# In-memory store for pending approval requests
+_pending_requests: dict[str, dict[str, Any]] = {}
+_decisions: dict[str, dict[str, Any]] = {}
+
+
+def create_owner_approval_request(change: dict[str, Any], architect_decision: dict[str, Any]) -> dict[str, Any]:
+    req = gate.create_request(change, architect_decision)
+    d = asdict(req)
+    _pending_requests[req.change_id] = d
+    return d
+
+
+def approve_request(change_id: str, decided_by: str = "owner", reason: str = "") -> dict[str, Any]:
+    req = _pending_requests.get(change_id)
+    if not req:
+        return {"error": f"no pending request for change_id={change_id}"}
+    decision = asdict(
+        ArchitectureOwnerApprovalRequest.__bases__[1].__new__(type("_ArchitectureOwnerDecision", (), {}))  # noqa
+    )
+    try:
+        from .architecture_owner_gate import ArchitectureOwnerDecision
+        dec = ArchitectureOwnerDecision(
+            request_id=req["request_id"],
+            change_id=change_id,
+            decision="approved",
+            decided_by=decided_by,
+            reason=reason,
+        )
+        d = asdict(dec)
+        _decisions[change_id] = d
+        # remove from pending
+        _pending_requests.pop(change_id, None)
+        return {"allowed": True, "decision": d}
+    except Exception as exc:
+        return {"error": str(exc)}
+
+
+def reject_request(change_id: str, decided_by: str = "owner", reason: str = "") -> dict[str, Any]:
+    from .architecture_owner_gate import ArchitectureOwnerDecision
+    req = _pending_requests.get(change_id)
+    if not req:
+        return {"error": f"no pending request for change_id={change_id}"}
+    try:
+        dec = ArchitectureOwnerDecision(
+            request_id=req["request_id"],
+            change_id=change_id,
+            decision="rejected",
+            decided_by=decided_by,
+            reason=reason,
+        )
+        d = asdict(dec)
+        _decisions[change_id] = d
+        _pending_requests.pop(change_id, None)
+        return {"allowed": False, "decision": d}
+    except Exception as exc:
+        return {"error": str(exc)}
+
+
+def get_owner_gate_overview() -> dict[str, Any]:
+    return {
+        "pending_requests": list(_pending_requests.values()),
+        "decisions": list(_decisions.values()),
+        "pending_count": len(_pending_requests),
+        "approved_count": sum(1 for d in _decisions.values() if d.get("decision") == "approved"),
+        "rejected_count": sum(1 for d in _decisions.values() if d.get("decision") == "rejected"),
+    }
@@ -0,0 +1,40 @@
+"""Independent architecture optimization agent."""
+
+from dataclasses import dataclass, asdict
+from datetime import datetime, timezone
+
+
+def utc_now():
+    return datetime.now(timezone.utc).isoformat()
+
+
+@dataclass(frozen=True)
+class ArchitectureReview:
+    agent: str
+    decision: str
+    findings: list[str]
+    created_at: str = utc_now()
+
+    def to_dict(self):
+        return asdict(self)
+
+
+class ArchitectureReviewAgent:
+    """Reviews architecture changes from system-wide optimal perspective."""
+
+    def review(self, proposal: dict) -> ArchitectureReview:
+        findings = [
+            "dependency_impact_checked",
+            "security_boundary_checked",
+            "scalability_checked",
+            "operability_checked",
+            "rollback_strategy_checked",
+        ]
+        return ArchitectureReview(
+            "architecture-optimization-agent",
+            "pass" if proposal.get("architecture_change") else "not_required",
+            findings,
+        )
+
+
+__all__ = ["ArchitectureReviewAgent", "ArchitectureReview"]
@@ -0,0 +1,103 @@
+"""Change acceptance governance for AIOS.
+
+Every feature change must provide matching tests. Every architecture change must
+be reviewed by an independent architecture agent. User-facing changes must be
+validated by an independent user acceptance agent.
+"""
+
+from __future__ import annotations
+
+from dataclasses import asdict, dataclass, field
+from datetime import datetime, timezone
+from typing import Any
+
+
+def utc_now() -> str:
+    return datetime.now(timezone.utc).isoformat()
+
+
+@dataclass(frozen=True)
+class ChangeDescriptor:
+    change_id: str
+    change_type: str
+    title: str
+    files_changed: list[str]
+    tests: list[str] = field(default_factory=list)
+    user_facing: bool = False
+    architecture_impact: bool = False
+    risk: str = "low"
+
+    def to_dict(self) -> dict[str, Any]:
+        return asdict(self)
+
+
+@dataclass(frozen=True)
+class AcceptanceFinding:
+    check: str
+    status: str
+    detail: str
+    required: bool = True
+
+    def passed(self) -> bool:
+        return self.status == "pass" or not self.required
+
+    def to_dict(self) -> dict[str, Any]:
+        return asdict(self)
+
+
+@dataclass(frozen=True)
+class AcceptanceDecision:
+    status: str
+    findings: list[AcceptanceFinding]
+    decided_at: str = field(default_factory=utc_now)
+
+    def to_dict(self) -> dict[str, Any]:
+        return {
+            "status": self.status,
+            "findings": [f.to_dict() for f in self.findings],
+            "decided_at": self.decided_at,
+        }
+
+
+class ChangeAcceptanceGate:
+    def evaluate(self, change: ChangeDescriptor) -> AcceptanceDecision:
+        findings: list[AcceptanceFinding] = []
+        findings.append(self._tests_required(change))
+        findings.append(self._user_acceptance_required(change))
+        findings.append(self._architecture_review_required(change))
+        findings.append(self._risk_gate(change))
+        status = "pass" if all(f.passed() for f in findings) else "blocked"
+        return AcceptanceDecision(status, findings)
+
+    def _tests_required(self, change: ChangeDescriptor) -> AcceptanceFinding:
+        if change.tests:
+            return AcceptanceFinding("tests_present", "pass", "matching tests declared")
+        return AcceptanceFinding("tests_present", "fail", "every feature change requires tests")
+
+    def _user_acceptance_required(self, change: ChangeDescriptor) -> AcceptanceFinding:
+        if not change.user_facing:
+            return AcceptanceFinding("user_acceptance", "pass", "not user-facing")
+        has_user_test = any("user" in t.lower() or "acceptance" in t.lower() for t in change.tests)
+        return AcceptanceFinding(
+            "user_acceptance",
+            "pass" if has_user_test else "fail",
+            "user-facing change requires user acceptance test",
+        )
+
+    def _architecture_review_required(self, change: ChangeDescriptor) -> AcceptanceFinding:
+        if not change.architecture_impact:
+            return AcceptanceFinding("architecture_review", "pass", "no architecture impact")
+        has_arch_test = any("arch" in t.lower() or "architecture" in t.lower() for t in change.tests)
+        return AcceptanceFinding(
+            "architecture_review",
+            "pass" if has_arch_test else "fail",
+            "architecture change requires independent architect review artifact/test",
+        )
+
+    def _risk_gate(self, change: ChangeDescriptor) -> AcceptanceFinding:
+        if change.risk in {"low", "medium", "high"}:
+            return AcceptanceFinding("risk_classified", "pass", f"risk={change.risk}")
+        return AcceptanceFinding("risk_classified", "fail", "risk must be low/medium/high")
+
+
+__all__ = ["ChangeDescriptor", "AcceptanceFinding", "AcceptanceDecision", "ChangeAcceptanceGate"]
@@ -0,0 +1,34 @@
+"""API for AIOS change governance dashboard."""
+
+from typing import Any
+
+try:
+    from fastapi import APIRouter
+except Exception:
+    APIRouter = None
+
+from .change_governance_dashboard import ChangeGovernanceDashboard
+
+
+dashboard = ChangeGovernanceDashboard()
+
+
+def get_change_governance_overview(changes: list[dict[str, Any]] | None = None):
+    return dashboard.overview(changes)
+
+
+def build_router():
+    if APIRouter is None:
+        return None
+    router = APIRouter(prefix="/admin/ai-evolution/governance", tags=["change-governance"])
+
+    @router.get("/overview", operation_id="getChangeGovernanceOverview")
+    def overview():
+        return get_change_governance_overview()
+
+    return router
+
+
+router = build_router()
+
+__all__ = ["router", "get_change_governance_overview"]
@@ -0,0 +1,93 @@
+"""Web UI dashboard model for AIOS change governance.
+
+Shows test gates, user acceptance, architecture review, risk gates and release
+blocking status for human operators and owner review.
+"""
+
+from __future__ import annotations
+
+from dataclasses import asdict, dataclass, field
+from datetime import datetime, timezone
+from typing import Any
+
+
+def utc_now() -> str:
+    return datetime.now(timezone.utc).isoformat()
+
+
+@dataclass(frozen=True)
+class GovernanceMetric:
+    key: str
+    title: str
+    value: Any
+    status: str = "ok"
+    detail: str = ""
+
+    def to_dict(self) -> dict[str, Any]:
+        return asdict(self)
+
+
+@dataclass(frozen=True)
+class GovernancePanel:
+    key: str
+    title: str
+    metrics: list[GovernanceMetric] = field(default_factory=list)
+
+    def to_dict(self) -> dict[str, Any]:
+        return {
+            "key": self.key,
+            "title": self.title,
+            "metrics": [m.to_dict() for m in self.metrics],
+        }
+
+
+class ChangeGovernanceDashboard:
+    def overview(self, changes: list[dict[str, Any]] | None = None) -> dict[str, Any]:
+        changes = changes or []
+        blocked = [c for c in changes if c.get("status") == "blocked"]
+        waiting_arch = [c for c in changes if c.get("architecture_review") == "required"]
+        waiting_user = [c for c in changes if c.get("user_acceptance") == "required"]
+        missing_tests = [c for c in changes if not c.get("tests")]
+        panels = [
+            GovernancePanel(
+                "test_gate",
+                "测试 Gate",
+                [
+                    GovernanceMetric("total_changes", "变更总数", len(changes)),
+                    GovernanceMetric("missing_tests", "缺少测试的变更", len(missing_tests), "warn" if missing_tests else "ok"),
+                ],
+            ),
+            GovernancePanel(
+                "user_acceptance",
+                "用户验收 Agent",
+                [
+                    GovernanceMetric("waiting_user_acceptance", "等待用户验收", len(waiting_user), "warn" if waiting_user else "ok"),
+                ],
+            ),
+            GovernancePanel(
+                "architecture_review",
+                "架构师 Agent",
+                [
+                    GovernanceMetric("waiting_architecture_review", "等待架构审查", len(waiting_arch), "warn" if waiting_arch else "ok"),
+                ],
+            ),
+            GovernancePanel(
+                "release_gate",
+                "发布 Gate",
+                [
+                    GovernanceMetric("blocked_changes", "阻塞发布的变更", len(blocked), "warn" if blocked else "ok"),
+                ],
+            ),
+        ]
+        return {
+            "module": "change-governance",
+            "route": "/admin/ai-evolution/governance",
+            "title": "AIOS 变更治理中心",
+            "status": "attention_required" if blocked or missing_tests or waiting_arch or waiting_user else "healthy",
+            "generated_at": utc_now(),
+            "panels": [p.to_dict() for p in panels],
+            "changes": changes,
+        }
+
+
+__all__ = ["GovernanceMetric", "GovernancePanel", "ChangeGovernanceDashboard"]
@@ -0,0 +1,36 @@
+"""Git repository location resolver with validation.
+
+Prevents hard-coded repository names from silently pointing to stale paths.
+"""
+
+from __future__ import annotations
+
+import os
+from pathlib import Path
+
+
+class GitRepoLocator:
+    def __init__(self, root: str | None = None):
+        self.root = Path(root or os.getenv("AIOS_GITOPS_ROOT", "/data/aios-gitops/runtime/gitea-bin/data/repositories"))
+
+    def candidates(self, owner: str = "aios", names: tuple[str, ...] = ("aios.git", "aios-v23.git")):
+        return [self.root / owner / name for name in names]
+
+    def locate(self, owner: str = "aios") -> dict:
+        found = [str(path) for path in self.candidates(owner) if path.exists()]
+        return {
+            "root": str(self.root),
+            "owner": owner,
+            "found": found,
+            "healthy": len(found) == 1,
+            "warning": "multiple repositories found" if len(found) > 1 else None,
+        }
+
+    def require(self, owner: str = "aios") -> Path:
+        result = self.locate(owner)
+        if not result["healthy"]:
+            raise RuntimeError(f"Git repo resolution failed: {result}")
+        return Path(result["found"][0])
+
+
+__all__ = ["GitRepoLocator"]
@@ -0,0 +1,550 @@
+#!/usr/bin/env python3
+"""Isolated Git worktree lifecycle for AIOS GitOps write requests.
+
+The long-running controller workspace is read-only and remains detached.  Each
+write request gets a short-lived detached worktree.  A branch-scoped lease and
+compare-and-swap ref update prevent concurrent writers from silently replacing
+one another.
+"""
+from __future__ import annotations
+
+import fcntl
+import hashlib
+import json
+import os
+import re
+import shutil
+import subprocess
+import threading
+import uuid
+from dataclasses import dataclass
+from pathlib import Path
+from typing import Any, Iterable
+
+
+ZERO_OID = "0" * 40
+BRANCH_RE = re.compile(r"^[A-Za-z0-9][A-Za-z0-9._/-]{0,179}$")
+
+
+class GitWorkspaceError(RuntimeError):
+    """Base error for isolated worktree operations."""
+
+
+class BranchLeaseBusy(GitWorkspaceError):
+    """Another request currently owns the target branch lease."""
+
+
+class BranchCheckedOut(GitWorkspaceError):
+    """The target branch is attached to a non-ephemeral worktree."""
+
+
+class ConcurrentBranchUpdate(GitWorkspaceError):
+    """The branch changed after the request acquired its lease."""
+
+
+@dataclass
+class WorkspaceLease:
+    branch: str
+    base: str
+    path: Path
+    start_oid: str
+    local_oid_before: str | None
+    remote_oid_before: str | None
+    lock_file: Any
+
+
+class GitOpsWorktreeManager:
+    """Manage one detached worktree per GitOps write request.
+
+    A manager instance is safe for a ``ThreadingHTTPServer``.  Request state is
+    stored in thread-local storage, while a filesystem lease serializes writers
+    targeting the same branch across processes.
+    """
+
+    def __init__(
+        self,
+        controller_workspace: str | Path,
+        worktree_root: str | Path,
+        *,
+        remote: str = "origin",
+        lock_root: str | Path | None = None,
+        author_name: str = "chatgpt-gitops-bot",
+        author_email: str = "chatgpt-gitops-bot@cnnetmkt.top",
+    ) -> None:
+        self.controller_workspace = Path(controller_workspace).resolve()
+        self.worktree_root = Path(worktree_root).resolve()
+        self.remote = remote
+        self.author_name = author_name
+        self.author_email = author_email
+        self._state = threading.local()
+
+        if not self.controller_workspace.exists():
+            raise GitWorkspaceError(
+                f"controller workspace does not exist: {self.controller_workspace}"
+            )
+
+        common = self._git(
+            ["rev-parse", "--path-format=absolute", "--git-common-dir"],
+            cwd=self.controller_workspace,
+        ).stdout.strip()
+        self.common_git_dir = Path(common).resolve()
+        self.worktree_root.mkdir(parents=True, exist_ok=True)
+
+        preferred_lock_root = Path(
+            lock_root
+            or os.getenv("AIOS_GITOPS_BRANCH_LOCK_ROOT", "/run/lock/aios-gitops")
+        )
+        try:
+            preferred_lock_root.mkdir(parents=True, exist_ok=True)
+            self.lock_root = preferred_lock_root.resolve()
+        except PermissionError:
+            fallback = self.worktree_root / ".locks"
+            fallback.mkdir(parents=True, exist_ok=True)
+            self.lock_root = fallback.resolve()
+
+    @staticmethod
+    def validate_branch(branch: str) -> str:
+        value = str(branch or "").strip()
+        invalid = (
+            not BRANCH_RE.fullmatch(value)
+            or value in {"main", "master"}
+            or value.startswith(("main/", "master/"))
+            or value.endswith(("/", ".", ".lock"))
+            or value.startswith(("/", "."))
+            or ".." in value
+            or "//" in value
+            or "@{" in value
+            or "\\" in value
+        )
+        if invalid:
+            raise ValueError(f"invalid or protected branch: {value!r}")
+        return value
+
+    @staticmethod
+    def validate_path(path: str) -> str:
+        raw = str(path or "").replace("\\", "/").strip()
+        candidate = Path(raw)
+        if (
+            not raw
+            or candidate.is_absolute()
+            or ".." in candidate.parts
+            or candidate.parts[0] == ".git"
+        ):
+            raise ValueError(f"invalid repository path: {path!r}")
+        return str(candidate)
+
+    def _git(
+        self,
+        args: Iterable[str],
+        *,
+        cwd: Path | None = None,
+        check: bool = True,
+    ) -> subprocess.CompletedProcess[str]:
+        command = ["git", *list(args)]
+        proc = subprocess.run(
+            command,
+            cwd=str(cwd or self.controller_workspace),
+            text=True,
+            capture_output=True,
+        )
+        if check and proc.returncode != 0:
+            message = proc.stderr.strip() or proc.stdout.strip() or "git command failed"
+            raise GitWorkspaceError(f"{message}; command={command!r}")
+        return proc
+
+    def _git_dir(
+        self, args: Iterable[str], *, check: bool = True
+    ) -> subprocess.CompletedProcess[str]:
+        return self._git(
+            [f"--git-dir={self.common_git_dir}", *list(args)],
+            cwd=self.controller_workspace,
+            check=check,
+        )
+
+    def _rev_parse(self, ref: str) -> str | None:
+        proc = self._git_dir(["rev-parse", "--verify", ref], check=False)
+        return proc.stdout.strip() if proc.returncode == 0 else None
+
+    def _remote_oid(self, branch: str) -> str | None:
+        proc = self._git(
+            ["ls-remote", "--heads", self.remote, f"refs/heads/{branch}"],
+            cwd=self.controller_workspace,
+            check=False,
+        )
+        if proc.returncode != 0 or not proc.stdout.strip():
+            return None
+        return proc.stdout.split()[0]
+
+    def _fetch_remote_tracking(self, branch: str) -> str | None:
+        remote_oid = self._remote_oid(branch)
+        if not remote_oid:
+            return None
+        destination = f"refs/remotes/{self.remote}/{branch}"
+        self._git(
+            [
+                "fetch",
+                "--no-tags",
+                self.remote,
+                f"+refs/heads/{branch}:{destination}",
+            ],
+            cwd=self.controller_workspace,
+        )
+        return remote_oid
+
+    def _checked_out_paths(self, branch: str) -> list[str]:
+        target_ref = f"refs/heads/{branch}"
+        proc = self._git_dir(["worktree", "list", "--porcelain"])
+        paths: list[str] = []
+        current_path: str | None = None
+        for line in proc.stdout.splitlines():
+            if line.startswith("worktree "):
+                current_path = line.removeprefix("worktree ").strip()
+            elif line.startswith("branch ") and line.removeprefix("branch ").strip() == target_ref:
+                if current_path:
+                    paths.append(current_path)
+        return paths
+
+    def _assert_branch_not_checked_out(self, branch: str) -> None:
+        paths = self._checked_out_paths(branch)
+        if paths:
+            raise BranchCheckedOut(
+                f"branch {branch!r} is checked out at: {', '.join(paths)}"
+            )
+
+    def _acquire_lock(self, branch: str):
+        digest = hashlib.sha256(branch.encode("utf-8")).hexdigest()
+        path = self.lock_root / f"{digest}.lock"
+        handle = path.open("a+", encoding="utf-8")
+        try:
+            fcntl.flock(handle.fileno(), fcntl.LOCK_EX | fcntl.LOCK_NB)
+        except BlockingIOError as exc:
+            handle.close()
+            raise BranchLeaseBusy(f"branch lease is busy: {branch}") from exc
+        handle.seek(0)
+        handle.truncate()
+        handle.write(json.dumps({"branch": branch, "pid": os.getpid()}))
+        handle.flush()
+        return handle
+
+    def _release_lock(self, handle: Any) -> None:
+        try:
+            fcntl.flock(handle.fileno(), fcntl.LOCK_UN)
+        finally:
+            handle.close()
+
+    def _current(self) -> WorkspaceLease | None:
+        return getattr(self._state, "lease", None)
+
+    def current_workspace(self) -> Path:
+        lease = self._current()
+        return lease.path if lease else self.controller_workspace
+
+    def begin(self, branch: str, base: str = "main", create: bool = False) -> str:
+        del create  # Existing branches are resumed; they are never destructively reset.
+        if self._current() is not None:
+            raise GitWorkspaceError("request already owns a worktree lease")
+
+        branch = self.validate_branch(branch)
+        base = str(base or "main").strip()
+        lock_file = self._acquire_lock(branch)
+        try:
+            self._assert_branch_not_checked_out(branch)
+            remote_oid = self._fetch_remote_tracking(branch)
+            local_oid = self._rev_parse(f"refs/heads/{branch}")
+
+            if remote_oid and local_oid and remote_oid != local_oid:
+                raise ConcurrentBranchUpdate(
+                    f"local and remote branch diverged before write: {branch}"
+                )
+
+            start_oid = remote_oid or local_oid
+            if not start_oid:
+                base_remote = self._fetch_remote_tracking(base)
+                base_local = self._rev_parse(f"refs/heads/{base}")
+                start_oid = base_remote or base_local or self._rev_parse(base)
+            if not start_oid:
+                raise GitWorkspaceError(f"cannot resolve base ref: {base}")
+
+            suffix = uuid.uuid4().hex[:12]
+            safe_branch = re.sub(r"[^A-Za-z0-9._-]+", "-", branch)[:96]
+            path = self.worktree_root / f"{safe_branch}-{suffix}"
+            self._git_dir(["worktree", "add", "--detach", str(path), start_oid])
+            self._state.lease = WorkspaceLease(
+                branch=branch,
+                base=base,
+                path=path,
+                start_oid=start_oid,
+                local_oid_before=local_oid,
+                remote_oid_before=remote_oid,
+                lock_file=lock_file,
+            )
+            return branch
+        except Exception:
+            self._release_lock(lock_file)
+            raise
+
+    def sh(self, args: list[str], check: bool = False) -> dict[str, Any]:
+        proc = self._git(args, cwd=self.current_workspace(), check=False)
+        if check and proc.returncode != 0:
+            message = proc.stderr.strip() or proc.stdout.strip() or "git command failed"
+            raise GitWorkspaceError(message)
+        return {
+            "code": proc.returncode,
+            "stdout": proc.stdout.strip(),
+            "stderr": proc.stderr.strip(),
+            "cmd": ["git", *args],
+        }
+
+    def write_file(self, path: str, content: str) -> str:
+        lease = self._current()
+        if lease is None:
+            raise GitWorkspaceError("write attempted without an active branch lease")
+        relative = self.validate_path(path)
+        target = lease.path / relative
+        target.parent.mkdir(parents=True, exist_ok=True)
+        target.write_text(content, encoding="utf-8")
+        return relative
+
+    def write_json(self, path: str, data: Any) -> str:
+        return self.write_file(
+            path, json.dumps(data, ensure_ascii=False, indent=2) + "\n"
+        )
+
+    def _update_local_branch(self, lease: WorkspaceLease, new_oid: str) -> None:
+        self._assert_branch_not_checked_out(lease.branch)
+        ref = f"refs/heads/{lease.branch}"
+        expected = lease.local_oid_before or ZERO_OID
+        proc = self._git_dir(["update-ref", ref, new_oid, expected], check=False)
+        if proc.returncode != 0:
+            raise ConcurrentBranchUpdate(
+                proc.stderr.strip()
+                or f"local branch changed concurrently: {lease.branch}"
+            )
+
+    def _rollback_local_branch(self, lease: WorkspaceLease, new_oid: str) -> None:
+        ref = f"refs/heads/{lease.branch}"
+        if lease.local_oid_before:
+            self._git_dir(
+                ["update-ref", ref, lease.local_oid_before, new_oid], check=False
+            )
+        else:
+            self._git_dir(["update-ref", "-d", ref, new_oid], check=False)
+
+    def _push_remote(self, lease: WorkspaceLease, new_oid: str) -> dict[str, Any]:
+        remote_url = self._git(
+            ["remote", "get-url", self.remote],
+            cwd=lease.path,
+            check=False,
+        )
+        if remote_url.returncode != 0:
+            return {"status": "synced", "target": "common_git_dir"}
+
+        command = ["push", "--porcelain"]
+        if lease.remote_oid_before:
+            command.append(
+                f"--force-with-lease=refs/heads/{lease.branch}:{lease.remote_oid_before}"
+            )
+        command.extend(
+            [self.remote, f"HEAD:refs/heads/{lease.branch}"]
+        )
+        proc = self._git(command, cwd=lease.path, check=False)
+        if proc.returncode != 0:
+            self._rollback_local_branch(lease, new_oid)
+            raise ConcurrentBranchUpdate(
+                proc.stderr.strip()
+                or proc.stdout.strip()
+                or f"remote branch changed concurrently: {lease.branch}"
+            )
+
+        tracking_ref = f"refs/remotes/{self.remote}/{lease.branch}"
+        tracking_expected = lease.remote_oid_before or ZERO_OID
+        self._git_dir(
+            ["update-ref", tracking_ref, new_oid, tracking_expected], check=False
+        )
+        return {
+            "status": "synced",
+            "target": self.remote,
+            "stdout": proc.stdout.strip(),
+            "stderr": proc.stderr.strip(),
+        }
+
+    def commit(self, paths: list[str] | str | None, message: str) -> dict[str, Any]:
+        lease = self._current()
+        if lease is None:
+            raise GitWorkspaceError("commit attempted without an active branch lease")
+
+        raw_paths = paths or ["."]
+        if isinstance(raw_paths, str):
+            raw_paths = [raw_paths]
+        safe_paths = [self.validate_path(p) if p != "." else "." for p in raw_paths]
+        self._git(["add", "--", *safe_paths], cwd=lease.path)
+        status = self._git(["status", "--porcelain"], cwd=lease.path).stdout.strip()
+        if not status:
+            result = {
+                "status": "no_changes",
+                "branch": lease.branch,
+                "rev": lease.start_oid[:12],
+                "sync": {"status": "unchanged"},
+                "workspace_mode": "detached",
+            }
+            self.finish(success=True)
+            return result
+
+        commit_proc = self._git(
+            [
+                "-c",
+                f"user.name={self.author_name}",
+                "-c",
+                f"user.email={self.author_email}",
+                "commit",
+                "-m",
+                message,
+            ],
+            cwd=lease.path,
+        )
+        new_oid = self._git(["rev-parse", "HEAD"], cwd=lease.path).stdout.strip()
+        self._update_local_branch(lease, new_oid)
+        sync = self._push_remote(lease, new_oid)
+        result = {
+            "status": "committed",
+            "branch": lease.branch,
+            "rev": new_oid[:12],
+            "commit": new_oid,
+            "git": {
+                "code": commit_proc.returncode,
+                "stdout": commit_proc.stdout.strip(),
+                "stderr": commit_proc.stderr.strip(),
+            },
+            "sync": sync,
+            "workspace_mode": "detached",
+        }
+        self.finish(success=True)
+        return result
+
+    def finish(self, *, success: bool) -> dict[str, Any] | None:
+        lease = self._current()
+        if lease is None:
+            return None
+        self._state.lease = None
+        retained = False
+        cleanup_error: str | None = None
+        try:
+            status = self._git(
+                ["status", "--porcelain"], cwd=lease.path, check=False
+            ).stdout.strip()
+            if success and not status:
+                proc = self._git_dir(
+                    ["worktree", "remove", str(lease.path)], check=False
+                )
+                if proc.returncode != 0:
+                    cleanup_error = proc.stderr.strip() or proc.stdout.strip()
+                    retained = True
+            else:
+                retained = lease.path.exists()
+        finally:
+            self._release_lock(lease.lock_file)
+            self._git_dir(["worktree", "prune", "--expire", "now"], check=False)
+        return {
+            "branch": lease.branch,
+            "worktree": str(lease.path),
+            "retained": retained,
+            "cleanup_error": cleanup_error,
+        }
+
+    def abort_current(self) -> dict[str, Any] | None:
+        """Release an unfinished request, retaining dirty evidence for review."""
+        lease = self._current()
+        if lease is None:
+            return None
+        status = self._git(
+            ["status", "--porcelain"], cwd=lease.path, check=False
+        ).stdout.strip()
+        return self.finish(success=not bool(status))
+
+    def sync_branch(self, branch: str) -> dict[str, Any]:
+        """Compatibility hook for the legacy API entrypoint."""
+        return {
+            "status": "managed_by_isolated_worktree",
+            "branch": self.validate_branch(branch),
+        }
+
+    def sync_tag(self, tag: str) -> dict[str, Any]:
+        proc = self._git(
+            ["push", self.remote, f"refs/tags/{tag}:refs/tags/{tag}"],
+            cwd=self.controller_workspace,
+            check=False,
+        )
+        if proc.returncode != 0:
+            raise GitWorkspaceError(proc.stderr.strip() or proc.stdout.strip())
+        return {
+            "status": "synced",
+            "target": self.remote,
+            "tag": tag,
+            "stdout": proc.stdout.strip(),
+            "stderr": proc.stderr.strip(),
+        }
+
+    def runtime_status(self) -> dict[str, Any]:
+        branch = self._git(
+            ["branch", "--show-current"], cwd=self.controller_workspace
+        ).stdout.strip()
+        head = self._git(
+            ["rev-parse", "HEAD"], cwd=self.controller_workspace
+        ).stdout.strip()
+        dirty = bool(
+            self._git(
+                ["status", "--porcelain"], cwd=self.controller_workspace
+            ).stdout.strip()
+        )
+        return {
+            "workspace_mode": "branch" if branch else "detached",
+            "workspace_branch": branch or None,
+            "workspace_ref": head,
+            "workspace_dirty": dirty,
+            "common_git_dir": str(self.common_git_dir),
+            "task_worktree_root": str(self.worktree_root),
+        }
+
+
+def remove_abandoned_clean_worktrees(
+    manager: GitOpsWorktreeManager, *, prefix: str | Path | None = None
+) -> list[str]:
+    """Remove clean detached worktrees below the managed root.
+
+    Dirty worktrees are deliberately retained.  This helper is intended for a
+    startup recovery job and never removes worktrees outside the configured
+    managed root.
+    """
+    root = Path(prefix or manager.worktree_root).resolve()
+    removed: list[str] = []
+    proc = manager._git_dir(["worktree", "list", "--porcelain"])
+    candidates: list[Path] = []
+    current: Path | None = None
+    detached = False
+    for line in [*proc.stdout.splitlines(), ""]:
+        if line.startswith("worktree "):
+            if current and detached:
+                candidates.append(current)
+            current = Path(line.removeprefix("worktree ").strip()).resolve()
+            detached = False
+        elif line == "detached":
+            detached = True
+        elif not line and current and detached:
+            candidates.append(current)
+            current = None
+            detached = False
+
+    for path in candidates:
+        if path == manager.controller_workspace or root not in path.parents:
+            continue
+        status = manager._git(["status", "--porcelain"], cwd=path, check=False)
+        if status.returncode == 0 and not status.stdout.strip():
+            removal = manager._git_dir(
+                ["worktree", "remove", str(path)], check=False
+            )
+            if removal.returncode == 0:
+                removed.append(str(path))
+            elif not path.exists():
+                shutil.rmtree(path, ignore_errors=True)
+    manager._git_dir(["worktree", "prune", "--expire", "now"], check=False)
+    return removed
@@ -0,0 +1,148 @@
+"""Bridge governance agents into the unified Agent Event Bus.
+
+This module emits normalized governance events for requirement changes,
+architecture reviews, user acceptance, test gates and version decisions. It uses
+duck typing so it can work with the unified AgentEventBus when available, while
+remaining testable without a running bus.
+"""
+
+from __future__ import annotations
+
+from dataclasses import asdict, dataclass, field
+from datetime import datetime, timezone
+from typing import Any, Protocol
+from uuid import uuid4
+
+
+def utc_now() -> str:
+    return datetime.now(timezone.utc).isoformat()
+
+
+class EventPublisher(Protocol):
+    def publish(
+        self,
+        kind: str,
+        source: str,
+        target: str,
+        summary: str,
+        payload: dict[str, Any],
+        correlation_id: str | None = None,
+    ) -> dict[str, Any]:
+        ...
+
+
+@dataclass(frozen=True)
+class GovernanceEventEnvelope:
+    id: str
+    kind: str
+    source: str
+    target: str
+    summary: str
+    payload: dict[str, Any]
+    correlation_id: str | None = None
+    created_at: str = field(default_factory=utc_now)
+
+    def to_dict(self) -> dict[str, Any]:
+        return asdict(self)
+
+
+class InMemoryGovernanceEventBus:
+    """Small fallback bus for tests and offline governance runs."""
+
+    def __init__(self) -> None:
+        self.events: list[dict[str, Any]] = []
+
+    def publish(
+        self,
+        kind: str,
+        source: str,
+        target: str,
+        summary: str,
+        payload: dict[str, Any],
+        correlation_id: str | None = None,
+    ) -> dict[str, Any]:
+        event = GovernanceEventEnvelope(
+            id=f"evt_{uuid4().hex[:12]}",
+            kind=kind,
+            source=source,
+            target=target,
+            summary=summary,
+            payload=payload,
+            correlation_id=correlation_id,
+        ).to_dict()
+        self.events.append(event)
+        return event
+
+
+class GovernanceEventBridge:
+    """Publishes governance lifecycle events to the Agent Event Bus."""
+
+    def __init__(self, bus: EventPublisher | None = None):
+        self.bus = bus or InMemoryGovernanceEventBus()
+
+    def requirement_changed(self, requirement: dict[str, Any]) -> dict[str, Any]:
+        return self.bus.publish(
+            kind="requirement_changed",
+            source=requirement.get("source", "unknown"),
+            target="architect-agent",
+            summary=requirement.get("requirement", "requirement changed"),
+            payload=requirement,
+            correlation_id=requirement.get("id"),
+        )
+
+    def architecture_review_started(self, proposal: dict[str, Any]) -> dict[str, Any]:
+        return self.bus.publish(
+            kind="architecture_review_started",
+            source="architect-agent",
+            target="architecture-review-agent",
+            summary=proposal.get("title", "architecture review started"),
+            payload=proposal,
+            correlation_id=proposal.get("change_id"),
+        )
+
+    def architecture_review_completed(self, decision: dict[str, Any]) -> dict[str, Any]:
+        return self.bus.publish(
+            kind="architecture_review_completed",
+            source="architecture-review-agent",
+            target="change-governance",
+            summary=f"architecture review {decision.get('status', 'completed')}",
+            payload=decision,
+            correlation_id=decision.get("change_id"),
+        )
+
+    def user_acceptance_completed(self, report: dict[str, Any]) -> dict[str, Any]:
+        return self.bus.publish(
+            kind="user_acceptance_completed",
+            source="user-acceptance-agent",
+            target="change-governance",
+            summary=f"user acceptance {report.get('decision', 'completed')}",
+            payload=report,
+            correlation_id=report.get("change_id"),
+        )
+
+    def test_gate_completed(self, result: dict[str, Any]) -> dict[str, Any]:
+        return self.bus.publish(
+            kind="test_gate_completed",
+            source="test-gate",
+            target="change-governance",
+            summary=f"test gate {result.get('status', 'completed')}",
+            payload=result,
+            correlation_id=result.get("change_id"),
+        )
+
+    def version_decision_created(self, decision: dict[str, Any]) -> dict[str, Any]:
+        return self.bus.publish(
+            kind="version_decision_created",
+            source="architect-agent",
+            target="owner-gate",
+            summary=f"version bump recommendation: {decision.get('version_bump', 'unknown')}",
+            payload=decision,
+            correlation_id=decision.get("change_id"),
+        )
+
+
+__all__ = [
+    "GovernanceEventEnvelope",
+    "InMemoryGovernanceEventBus",
+    "GovernanceEventBridge",
+]
@@ -0,0 +1,32 @@
+"""API for governance timeline visualization."""
+
+try:
+    from fastapi import APIRouter
+except Exception:
+    APIRouter = None
+
+from .governance_timeline_dashboard import GovernanceTimelineDashboard
+
+
+dashboard = GovernanceTimelineDashboard()
+
+
+def get_governance_timeline(events=None):
+    return dashboard.build(events)
+
+
+def build_router():
+    if APIRouter is None:
+        return None
+    router = APIRouter(prefix="/admin/ai-evolution/governance", tags=["governance-timeline"])
+
+    @router.get("/timeline", operation_id="getGovernanceTimeline")
+    def timeline():
+        return get_governance_timeline()
+
+    return router
+
+
+router = build_router()
+
+__all__ = ["router", "get_governance_timeline"]
@@ -0,0 +1,106 @@
+"""Governance timeline dashboard.
+
+Turns governance events into a human-readable flow for Web UI so operators can
+see why a change is blocked or ready for promotion.
+"""
+
+from __future__ import annotations
+
+from dataclasses import asdict, dataclass, field
+from datetime import datetime, timezone
+from typing import Any
+
+
+def utc_now() -> str:
+    return datetime.now(timezone.utc).isoformat()
+
+
+EVENT_ORDER = {
+    "requirement_changed": 10,
+    "architecture_review_started": 20,
+    "architecture_review_completed": 30,
+    "test_gate_completed": 40,
+    "user_acceptance_completed": 50,
+    "version_decision_created": 60,
+    "owner_gate_completed": 70,
+    "promotion_completed": 80,
+}
+
+
+@dataclass(frozen=True)
+class GovernanceTimelineStep:
+    id: str
+    kind: str
+    title: str
+    actor: str
+    status: str
+    timestamp: str
+    correlation_id: str | None = None
+    detail: dict[str, Any] = field(default_factory=dict)
+
+    def to_dict(self) -> dict[str, Any]:
+        return asdict(self)
+
+
+class GovernanceTimelineDashboard:
+    def build(self, events: list[dict[str, Any]] | None = None) -> dict[str, Any]:
+        events = events or []
+        steps = [self._event_to_step(event) for event in events]
+        steps.sort(key=lambda step: (EVENT_ORDER.get(step.kind, 999), step.timestamp))
+        blockers = [step for step in steps if step.status in {"blocked", "fail", "attention_required"}]
+        return {
+            "module": "governance-timeline",
+            "route": "/admin/ai-evolution/governance/timeline",
+            "title": "治理事件时间线",
+            "status": "attention_required" if blockers else "healthy",
+            "generated_at": utc_now(),
+            "steps": [step.to_dict() for step in steps],
+            "blockers": [step.to_dict() for step in blockers],
+            "flow": self._flow_summary(steps),
+        }
+
+    def _event_to_step(self, event: dict[str, Any]) -> GovernanceTimelineStep:
+        payload = event.get("payload", {})
+        kind = event.get("kind", "unknown")
+        return GovernanceTimelineStep(
+            id=event.get("id", f"event-{len(str(event))}"),
+            kind=kind,
+            title=self._title(kind, payload),
+            actor=event.get("source", payload.get("actor", "unknown")),
+            status=payload.get("status") or payload.get("decision") or event.get("status", "ok"),
+            timestamp=event.get("created_at") or payload.get("created_at") or utc_now(),
+            correlation_id=event.get("correlation_id") or payload.get("change_id") or payload.get("id"),
+            detail=payload,
+        )
+
+    def _title(self, kind: str, payload: dict[str, Any]) -> str:
+        titles = {
+            "requirement_changed": "需求变化",
+            "architecture_review_started": "架构审查开始",
+            "architecture_review_completed": "架构审查完成",
+            "test_gate_completed": "测试 Gate 完成",
+            "user_acceptance_completed": "用户验收完成",
+            "version_decision_created": "版本建议产生",
+            "owner_gate_completed": "Owner Gate 完成",
+            "promotion_completed": "Promotion 完成",
+        }
+        return payload.get("title") or payload.get("summary") or titles.get(kind, kind)
+
+    def _flow_summary(self, steps: list[GovernanceTimelineStep]) -> dict[str, Any]:
+        present = {step.kind for step in steps}
+        required = [
+            "requirement_changed",
+            "architecture_review_completed",
+            "test_gate_completed",
+            "user_acceptance_completed",
+            "version_decision_created",
+        ]
+        missing = [kind for kind in required if kind not in present]
+        return {
+            "required_steps": required,
+            "missing_steps": missing,
+            "ready_for_owner_gate": not missing,
+        }
+
+
+__all__ = ["GovernanceTimelineStep", "GovernanceTimelineDashboard"]
@@ -0,0 +1,59 @@
+"""Independent acceptance agents.
+
+Simulates independent roles: user representative and architecture reviewer.
+They provide acceptance evidence, not implementation decisions.
+"""
+
+from dataclasses import dataclass, asdict
+from datetime import datetime, timezone
+from typing import Any
+
+
+def utc_now():
+    return datetime.now(timezone.utc).isoformat()
+
+
+@dataclass(frozen=True)
+class AcceptanceReport:
+    agent: str
+    role: str
+    decision: str
+    evidence: list[str]
+    created_at: str = utc_now()
+
+    def to_dict(self):
+        return asdict(self)
+
+
+class UserAcceptanceAgent:
+    def review(self, change: dict[str, Any]) -> AcceptanceReport:
+        checks = [
+            "user_requirement_match",
+            "user_workflow_validation",
+            "permission_visibility_check",
+        ]
+        return AcceptanceReport(
+            "user-acceptance-agent",
+            "user_representative",
+            "pass" if change.get("user_facing") else "not_required",
+            checks,
+        )
+
+
+class ArchitectureAcceptanceAgent:
+    def review(self, change: dict[str, Any]) -> AcceptanceReport:
+        checks = [
+            "architecture_consistency",
+            "scalability_review",
+            "security_boundary_review",
+            "operability_review",
+        ]
+        return AcceptanceReport(
+            "architecture-review-agent",
+            "system_architect",
+            "pass" if change.get("architecture_impact") else "not_required",
+            checks,
+        )
+
+
+__all__ = ["AcceptanceReport", "UserAcceptanceAgent", "ArchitectureAcceptanceAgent"]
@@ -0,0 +1,92 @@
+"""Agent adapters for AIOS Search Gateway.
+
+All runtime agents should call SearchGatewayAgentAdapter instead of performing
+ad-hoc network search. This gives consistent provider fallback, profile
+resolution, evidence capture, audit and event visibility.
+"""
+
+from __future__ import annotations
+
+from dataclasses import asdict, dataclass, field
+from typing import Any
+
+from .search_gateway_orchestrator import SearchGatewayOrchestrator
+
+
+@dataclass(frozen=True)
+class AgentSearchContext:
+    agent: str
+    user_id: str | None = None
+    group_id: str | None = None
+    role: str = "user"
+    group_profile: dict[str, Any] | None = None
+    user_profile: dict[str, Any] | None = None
+    session_profile: dict[str, Any] | None = None
+    authorized_sources: list[str] = field(default_factory=list)
+
+    def to_dict(self) -> dict[str, Any]:
+        return asdict(self)
+
+
+class SearchGatewayAgentAdapter:
+    def __init__(self, orchestrator: SearchGatewayOrchestrator | None = None):
+        self.orchestrator = orchestrator or SearchGatewayOrchestrator()
+
+    def search(
+        self,
+        query: str,
+        context: AgentSearchContext,
+        intent: str = "general",
+        region: str | None = None,
+        time_range: str | None = None,
+        data_type: str | None = None,
+        domain_profiles: list[str] | None = None,
+        user_scope: str = "public",
+        correlation_id: str | None = None,
+    ) -> dict[str, Any]:
+        payload = {
+            "query": query,
+            "requester": context.agent,
+            "intent": intent,
+            "region": region,
+            "time_range": time_range,
+            "data_type": data_type,
+            "domain_profiles": domain_profiles or [],
+            "user_scope": user_scope,
+            "authorized_sources": context.authorized_sources,
+            "group_profile": context.group_profile,
+            "user_profile": context.user_profile,
+            "session_profile": context.session_profile,
+            "correlation_id": correlation_id,
+        }
+        return self.orchestrator.execute(payload)
+
+
+class OpenClawSearchClient(SearchGatewayAgentAdapter):
+    def search_public(self, query: str, context: AgentSearchContext, **kwargs) -> dict[str, Any]:
+        return self.search(query, context, user_scope="public", **kwargs)
+
+
+class HermesSearchClient(SearchGatewayAgentAdapter):
+    def delegate_search(self, query: str, context: AgentSearchContext, **kwargs) -> dict[str, Any]:
+        return self.search(query, context, **kwargs)
+
+
+class CodexSearchClient(SearchGatewayAgentAdapter):
+    def technical_search(self, query: str, context: AgentSearchContext, **kwargs) -> dict[str, Any]:
+        return self.search(query, context, intent=kwargs.pop("intent", "software_library"), **kwargs)
+
+
+class CustomGPTSearchClient(SearchGatewayAgentAdapter):
+    def user_search(self, query: str, context: AgentSearchContext, **kwargs) -> dict[str, Any]:
+        return self.search(query, context, **kwargs)
+
+
+__all__ = [
+    "AgentSearchContext",
+    "SearchGatewayAgentAdapter",
+    "OpenClawSearchClient",
+    "HermesSearchClient",
+    "CodexSearchClient",
+    "CustomGPTSearchClient",
+]
@@ -0,0 +1,73 @@
+"""Architect governance integration for Search Gateway.
+
+Search Gateway is a shared platform capability. The Architect Agent periodically
+reviews provider health, external search/browser capabilities, cost, security and
+privacy impact, then emits C/R/V recommendations.
+"""
+
+from __future__ import annotations
+
+from dataclasses import asdict, dataclass, field
+from datetime import datetime, timezone
+from typing import Any
+
+
+def utc_now() -> str:
+    return datetime.now(timezone.utc).isoformat()
+
+
+@dataclass(frozen=True)
+class SearchArchitectureFinding:
+    finding_id: str
+    title: str
+    impact: str
+    recommendation: str
+    evidence: list[str] = field(default_factory=list)
+    created_at: str = field(default_factory=utc_now)
+
+    def to_dict(self) -> dict[str, Any]:
+        return asdict(self)
+
+
+@dataclass(frozen=True)
+class SearchArchitectureDecision:
+    status: str
+    version_bump: str
+    reason: str
+    findings: list[SearchArchitectureFinding]
+    owner_gate_required: bool
+    created_at: str = field(default_factory=utc_now)
+
+    def to_dict(self) -> dict[str, Any]:
+        return {
+            "status": self.status,
+            "version_bump": self.version_bump,
+            "reason": self.reason,
+            "findings": [f.to_dict() for f in self.findings],
+            "owner_gate_required": self.owner_gate_required,
+            "created_at": self.created_at,
+        }
+
+
+class SearchArchitectIntegration:
+    def review(self, signals: dict[str, Any]) -> SearchArchitectureDecision:
+        findings: list[SearchArchitectureFinding] = []
+        if signals.get("provider_breaking_change"):
+            findings.append(SearchArchitectureFinding("search-provider-breaking", "Provider breaking change", "breaking", "owner_review"))
+        if signals.get("new_search_capability"):
+            findings.append(SearchArchitectureFinding("search-new-capability", "New search capability available", "architecture", "evaluate_adoption"))
+        if signals.get("cost_increase"):
+            findings.append(SearchArchitectureFinding("search-cost", "Search provider cost increase", "minor", "optimize_provider_order"))
+        if signals.get("security_advisory"):
+            findings.append(SearchArchitectureFinding("search-security", "Search security advisory", "breaking", "owner_review"))
+
+        if any(f.impact == "breaking" for f in findings):
+            return SearchArchitectureDecision("action_required", "V", "breaking search ecosystem/security change", findings, True)
+        if any(f.impact == "architecture" for f in findings):
+            return SearchArchitectureDecision("optimization_recommended", "R", "new search capability may improve architecture", findings, True)
+        if findings:
+            return SearchArchitectureDecision("minor_change_recommended", "C", "compatible provider optimization", findings, False)
+        return SearchArchitectureDecision("no_action", "none", "no relevant search architecture change", findings, False)
+
+
+__all__ = ["SearchArchitectureFinding", "SearchArchitectureDecision", "SearchArchitectIntegration"]
@@ -0,0 +1,27 @@
+"""Audit storage for Search Gateway executions."""
+
+from __future__ import annotations
+
+import json
+from pathlib import Path
+from typing import Any
+
+
+class SearchAuditStore:
+    def __init__(self, root: str = "runtime/search-gateway/audit"):
+        self.root = Path(root)
+        self.root.mkdir(parents=True, exist_ok=True)
+
+    def append(self, record: dict[str, Any]) -> str:
+        path = self.root / f"{record.get('request_id', 'search')}.json"
+        path.write_text(json.dumps(record, ensure_ascii=False, indent=2), encoding="utf-8")
+        return str(path)
+
+    def list_recent(self) -> list[dict[str, Any]]:
+        result = []
+        for path in sorted(self.root.glob("*.json")):
+            result.append(json.loads(path.read_text(encoding="utf-8")))
+        return result[-50:]
+
+
+__all__ = ["SearchAuditStore"]
@@ -0,0 +1,118 @@
+"""Search Gateway event bridge.
+
+Publishes normalized search lifecycle events into the AIOS Agent Event Bus when a
+bus is available, with an in-memory fallback for tests and offline runs.
+"""
+
+from __future__ import annotations
+
+from dataclasses import asdict, dataclass, field
+from datetime import datetime, timezone
+from typing import Any, Protocol
+from uuid import uuid4
+
+
+def utc_now() -> str:
+    return datetime.now(timezone.utc).isoformat()
+
+
+class SearchEventPublisher(Protocol):
+    def publish(
+        self,
+        kind: str,
+        source: str,
+        target: str,
+        summary: str,
+        payload: dict[str, Any],
+        correlation_id: str | None = None,
+    ) -> dict[str, Any]:
+        ...
+
+
+@dataclass(frozen=True)
+class SearchEvent:
+    id: str
+    kind: str
+    source: str
+    target: str
+    summary: str
+    payload: dict[str, Any]
+    correlation_id: str | None = None
+    created_at: str = field(default_factory=utc_now)
+
+    def to_dict(self) -> dict[str, Any]:
+        return asdict(self)
+
+
+class InMemorySearchEventBus:
+    def __init__(self):
+        self.events: list[dict[str, Any]] = []
+
+    def publish(
+        self,
+        kind: str,
+        source: str,
+        target: str,
+        summary: str,
+        payload: dict[str, Any],
+        correlation_id: str | None = None,
+    ) -> dict[str, Any]:
+        event = SearchEvent(
+            id=f"search_evt_{uuid4().hex[:12]}",
+            kind=kind,
+            source=source,
+            target=target,
+            summary=summary,
+            payload=payload,
+            correlation_id=correlation_id,
+        ).to_dict()
+        self.events.append(event)
+        return event
+
+
+class SearchEventBridge:
+    def __init__(self, bus: SearchEventPublisher | None = None):
+        self.bus = bus or InMemorySearchEventBus()
+
+    def search_started(self, request: dict[str, Any]) -> dict[str, Any]:
+        return self.bus.publish(
+            kind="search_started",
+            source=request.get("requester", "unknown"),
+            target="search-gateway",
+            summary=request.get("query", "search started"),
+            payload=request,
+            correlation_id=request.get("correlation_id"),
+        )
+
+    def provider_attempted(self, request_id: str, attempt: dict[str, Any]) -> dict[str, Any]:
+        return self.bus.publish(
+            kind="search_provider_attempted",
+            source="search-gateway",
+            target=attempt.get("provider", "unknown"),
+            summary=f"provider {attempt.get('provider', 'unknown')} {attempt.get('status', 'attempted')}",
+            payload=attempt,
+            correlation_id=request_id,
+        )
+
+    def fallback_triggered(self, request_id: str, reason: str) -> dict[str, Any]:
+        return self.bus.publish(
+            kind="search_fallback_triggered",
+            source="search-gateway",
+            target="search-provider-router",
+            summary=reason,
+            payload={"reason": reason},
+            correlation_id=request_id,
+        )
+
+    def search_completed(self, response: dict[str, Any]) -> dict[str, Any]:
+        return self.bus.publish(
+            kind="search_completed",
+            source="search-gateway",
+            target=response.get("requester", "requester"),
+            summary=f"search {response.get('status', 'completed')}",
+            payload=response,
+            correlation_id=response.get("request_id"),
+        )
+
+
+__all__ = ["SearchEvent", "InMemorySearchEventBus", "SearchEventBridge"]
@@ -0,0 +1,252 @@
+"""Governed general-purpose Search Gateway for AIOS.
+
+The gateway coordinates query planning, provider routing, fallback, evidence
+normalization and compliance checks. It is provider-agnostic: runtime workers plug
+in public or explicitly authorized search providers.
+
+Search is a shared capability. Groups and users may customize provider order,
+trusted domains, query expansion hints and domain profiles, but customization may
+not override hard authorization, privacy, safety or audit rules.
+"""
+
+from __future__ import annotations
+
+from dataclasses import asdict, dataclass, field
+from datetime import datetime, timezone
+from typing import Any, Protocol
+from uuid import uuid4
+
+from .search_profile_resolver import DomainQueryProfileRegistry, SearchProfile, SearchProfileResolver
+
+
+def utc_now() -> str:
+    return datetime.now(timezone.utc).isoformat()
+
+
+@dataclass(frozen=True)
+class SearchRequest:
+    query: str
+    requester: str
+    intent: str = "general"
+    region: str | None = None
+    time_range: str | None = None
+    data_type: str | None = None
+    user_scope: str = "public"
+    authorized_sources: list[str] = field(default_factory=list)
+    domain_profiles: list[str] = field(default_factory=list)
+    group_profile: dict[str, Any] | None = None
+    user_profile: dict[str, Any] | None = None
+    session_profile: dict[str, Any] | None = None
+    correlation_id: str | None = None
+
+    def to_dict(self) -> dict[str, Any]:
+        return asdict(self)
+
+
+@dataclass(frozen=True)
+class SearchEvidence:
+    title: str
+    url: str | None
+    source: str
+    snippet: str
+    evidence_type: str = "web"
+    confidence: str = "medium"
+    retrieved_at: str = field(default_factory=utc_now)
+
+    def to_dict(self) -> dict[str, Any]:
+        return asdict(self)
+
+
+@dataclass(frozen=True)
+class SearchProviderAttempt:
+    provider: str
+    status: str
+    result_count: int = 0
+    reason: str = ""
+
+    def to_dict(self) -> dict[str, Any]:
+        return asdict(self)
+
+
+@dataclass(frozen=True)
+class SearchResponse:
+    request_id: str
+    status: str
+    answer: str
+    evidence: list[SearchEvidence]
+    attempts: list[SearchProviderAttempt]
+    confidence: str
+    expanded_queries: list[str] = field(default_factory=list)
+    resolved_profile: dict[str, Any] | None = None
+    need_clarification: list[str] = field(default_factory=list)
+    blocked_reasons: list[str] = field(default_factory=list)
+    created_at: str = field(default_factory=utc_now)
+
+    def to_dict(self) -> dict[str, Any]:
+        return {
+            "request_id": self.request_id,
+            "status": self.status,
+            "answer": self.answer,
+            "evidence": [e.to_dict() for e in self.evidence],
+            "attempts": [a.to_dict() for a in self.attempts],
+            "confidence": self.confidence,
+            "expanded_queries": self.expanded_queries,
+            "resolved_profile": self.resolved_profile,
+            "need_clarification": self.need_clarification,
+            "blocked_reasons": self.blocked_reasons,
+            "created_at": self.created_at,
+        }
+
+
+class SearchProvider(Protocol):
+    name: str
+
+    def search(self, request: SearchRequest, queries: list[str]) -> tuple[list[SearchEvidence], SearchProviderAttempt]:
+        ...
+
+
+class QueryPlanner:
+    def __init__(self, domain_registry: DomainQueryProfileRegistry | None = None):
+        self.domain_registry = domain_registry or DomainQueryProfileRegistry()
+
+    def expand(self, request: SearchRequest, profile: SearchProfile) -> list[str]:
+        base = request.query.strip()
+        queries = [base]
+        if request.region:
+            queries.append(f"{request.region} {base}")
+        if request.time_range:
+            queries.append(f"{base} {request.time_range}")
+        domain_profiles = list(dict.fromkeys([request.intent] + request.domain_profiles + profile.domain_profiles))
+        hints = list(profile.query_expansion_hints) + self.domain_registry.hints_for(domain_profiles)
+        for hint in hints:
+            queries.append(f"{base} {hint}")
+        for domain in profile.trusted_domains:
+            queries.append(f"site:{domain} {base}")
+        return list(dict.fromkeys([q for q in queries if q]))
+
+
+class SearchPolicyGuard:
+    """Blocks non-public or unauthorized collection patterns."""
+
+    FORBIDDEN_TERMS = (
+        "绕过登录",
+        "绕过验证码",
+        "验证码绕过",
+        "破解登录",
+        "个人成绩",
+        "学生隐私",
+        "受保护平台批量爬取",
+    )
+
+    def evaluate(self, request: SearchRequest, profile: SearchProfile) -> list[str]:
+        q = request.query.lower()
+        reasons: list[str] = []
+        for term in self.FORBIDDEN_TERMS:
+            if term.lower() in q:
+                reasons.append(f"blocked_term:{term}")
+        if request.user_scope not in {"public", "authorized_user_file", "authorized_link", "internal_archive"}:
+            reasons.append(f"unsupported_scope:{request.user_scope}")
+        if request.data_type in {"personal_grade", "student_private_data", "private_person_data"} and request.user_scope != "authorized_user_file":
+            reasons.append("private_data_requires_user_authorized_file")
+        if not profile.audit_required:
+            reasons.append("audit_cannot_be_disabled")
+        if not profile.fallback_required:
+            reasons.append("fallback_cannot_be_disabled")
+        return reasons
+
+
+class SearchGateway:
+    def __init__(self, providers: list[SearchProvider] | None = None):
+        self.providers = providers or []
+        self.profile_resolver = SearchProfileResolver()
+        self.planner = QueryPlanner()
+        self.guard = SearchPolicyGuard()
+
+    def search(self, request: SearchRequest) -> SearchResponse:
+        request_id = request.correlation_id or f"search_{uuid4().hex[:12]}"
+        profile = self.profile_resolver.resolve(request.group_profile, request.user_profile, request.session_profile)
+        expanded_queries = self.planner.expand(request, profile)
+        blocked = self.guard.evaluate(request, profile)
+        if blocked:
+            return SearchResponse(
+                request_id=request_id,
+                status="blocked",
+                answer="该搜索请求涉及未授权访问、验证码/登录绕过、受保护平台批量抓取或个人隐私数据，已阻断。可改为搜索公开网页、官方公告、公开文档，或处理用户授权上传的文件/链接。",
+                evidence=[],
+                attempts=[],
+                confidence="high",
+                expanded_queries=expanded_queries,
+                resolved_profile=profile.to_dict(),
+                blocked_reasons=blocked,
+            )
+
+        evidence: list[SearchEvidence] = []
+        attempts: list[SearchProviderAttempt] = []
+        providers = self._ordered_providers(profile)
+        for provider in providers:
+            try:
+                provider_evidence, attempt = provider.search(request, expanded_queries)
+                evidence.extend(provider_evidence)
+                attempts.append(attempt)
+                if provider_evidence:
+                    break
+            except Exception as exc:
+                attempts.append(SearchProviderAttempt(provider=getattr(provider, "name", "unknown"), status="failed", reason=str(exc)))
+
+        if evidence:
+            return SearchResponse(
+                request_id=request_id,
+                status="found",
+                answer="已找到公开或授权来源结果。",
+                evidence=evidence,
+                attempts=attempts,
+                confidence=self._confidence(evidence),
+                expanded_queries=expanded_queries,
+                resolved_profile=profile.to_dict(),
+            )
+
+        return SearchResponse(
+            request_id=request_id,
+            status="not_found",
+            answer="未找到足够可靠的公开或授权结果；已记录 provider 尝试和失败原因。",
+            evidence=[],
+            attempts=attempts,
+            confidence="low",
+            expanded_queries=expanded_queries,
+            resolved_profile=profile.to_dict(),
+            need_clarification=self._clarification(request),
+        )
+
+    def _ordered_providers(self, profile: SearchProfile) -> list[SearchProvider]:
+        if not self.providers:
+            return []
+        by_name = {provider.name: provider for provider in self.providers}
+        ordered = [by_name[name] for name in profile.provider_order if name in by_name]
+        remaining = [provider for provider in self.providers if provider.name not in {p.name for p in ordered}]
+        return ordered + remaining
+
+    def _confidence(self, evidence: list[SearchEvidence]) -> str:
+        if any(e.confidence == "high" for e in evidence):
+            return "high"
+        if evidence:
+            return "medium"
+        return "low"
+
+    def _clarification(self, request: SearchRequest) -> list[str]:
+        if request.intent == "education_exam_data":
+            return ["请补充学校名称、联考名称、年级、科目，以及需要试卷/答案还是成绩统计。"]
+        if request.intent in {"company_research", "policy_regulation", "software_library", "product_research"}:
+            return ["请补充主体名称、地区/版本、时间范围、需要的数据类型或可信来源范围。"]
+        return ["请补充更具体的主体、平台、地区、时间、数据类型或授权文件/链接。"]
+
+
+__all__ = [
+    "SearchRequest",
+    "SearchEvidence",
+    "SearchProviderAttempt",
+    "SearchResponse",
+    "SearchProvider",
+    "QueryPlanner",
+    "SearchPolicyGuard",
+    "SearchGateway",
+]
@@ -0,0 +1,65 @@
+"""API endpoints for AIOS Search Gateway."""
+
+from typing import Any
+
+try:
+    from fastapi import APIRouter
+except Exception:
+    APIRouter = None
+
+from .search_gateway import SearchGateway, SearchRequest
+from .search_gateway_dashboard import SearchGatewayDashboard
+
+
+gateway = SearchGateway()
+dashboard = SearchGatewayDashboard()
+
+
+def search_gateway_request(payload: dict[str, Any]) -> dict[str, Any]:
+    request = SearchRequest(**payload)
+    return gateway.search(request).to_dict()
+
+
+def get_search_gateway_status() -> dict[str, Any]:
+    return {
+        "module": "search-gateway",
+        "status": "ready",
+        "providers_configured": len(gateway.providers),
+        "policy_guard": "enabled",
+        "profile_customization": "enabled",
+        "fallback_required_when_provider_unavailable": True,
+    }
+
+
+def get_search_gateway_overview() -> dict[str, Any]:
+    return dashboard.overview()
+
+
+def build_router():
+    if APIRouter is None:
+        return None
+    router = APIRouter(prefix="/admin/ai-evolution/search", tags=["search-gateway"])
+
+    @router.get("/status", operation_id="getSearchGatewayStatus")
+    def status():
+        return get_search_gateway_status()
+
+    @router.get("/overview", operation_id="getSearchGatewayOverview")
+    def overview():
+        return get_search_gateway_overview()
+
+    @router.post("/query", operation_id="runSearchGatewayQuery")
+    def query(payload: dict[str, Any]):
+        return search_gateway_request(payload)
+
+    return router
+
+
+router = build_router()
+
+__all__ = [
+    "router",
+    "search_gateway_request",
+    "get_search_gateway_status",
+    "get_search_gateway_overview",
+]
@@ -0,0 +1,85 @@
+"""Search Gateway Web UI dashboard model."""
+
+from __future__ import annotations
+
+from dataclasses import asdict, dataclass, field
+from datetime import datetime, timezone
+from typing import Any
+
+
+def utc_now() -> str:
+    return datetime.now(timezone.utc).isoformat()
+
+
+@dataclass(frozen=True)
+class SearchDashboardMetric:
+    key: str
+    title: str
+    value: Any
+    status: str = "ok"
+    detail: str = ""
+
+    def to_dict(self) -> dict[str, Any]:
+        return asdict(self)
+
+
+@dataclass(frozen=True)
+class SearchDashboardPanel:
+    key: str
+    title: str
+    metrics: list[SearchDashboardMetric] = field(default_factory=list)
+
+    def to_dict(self) -> dict[str, Any]:
+        return {
+            "key": self.key,
+            "title": self.title,
+            "metrics": [m.to_dict() for m in self.metrics],
+        }
+
+
+class SearchGatewayDashboard:
+    def overview(self, search_runs: list[dict[str, Any]] | None = None, providers: list[dict[str, Any]] | None = None) -> dict[str, Any]:
+        search_runs = search_runs or []
+        providers = providers or []
+        blocked = [r for r in search_runs if r.get("status") == "blocked"]
+        not_found = [r for r in search_runs if r.get("status") == "not_found"]
+        provider_failures = [p for p in providers if p.get("status") == "failed"]
+        panels = [
+            SearchDashboardPanel(
+                "search_health",
+                "搜索健康度",
+                [
+                    SearchDashboardMetric("total_runs", "搜索请求数", len(search_runs)),
+                    SearchDashboardMetric("not_found", "未找到结果", len(not_found), "warn" if not_found else "ok"),
+                    SearchDashboardMetric("blocked", "合规阻断", len(blocked), "warn" if blocked else "ok"),
+                ],
+            ),
+            SearchDashboardPanel(
+                "provider_health",
+                "Provider 可用性",
+                [
+                    SearchDashboardMetric("providers", "Provider 数量", len(providers)),
+                    SearchDashboardMetric("provider_failures", "Provider 失败", len(provider_failures), "warn" if provider_failures else "ok"),
+                ],
+            ),
+            SearchDashboardPanel(
+                "evidence_quality",
+                "证据质量",
+                [
+                    SearchDashboardMetric("with_evidence", "有证据的回答", len([r for r in search_runs if r.get("evidence")])) ,
+                    SearchDashboardMetric("fallback_required", "要求多源降级", True),
+                ],
+            ),
+        ]
+        return {
+            "module": "search-gateway",
+            "route": "/admin/ai-evolution/search",
+            "title": "AIOS 联网搜索治理中心",
+            "status": "attention_required" if provider_failures else "healthy",
+            "generated_at": utc_now(),
+            "panels": [p.to_dict() for p in panels],
+            "recent_runs": search_runs[-20:],
+        }
+
+
+__all__ = ["SearchDashboardMetric", "SearchDashboardPanel", "SearchGatewayDashboard"]
@@ -0,0 +1,41 @@
+"""Runtime orchestrator joining Search Gateway, registry, audit and events."""
+
+from __future__ import annotations
+
+from typing import Any
+
+from .search_audit_store import SearchAuditStore
+from .search_event_bridge import SearchEventBridge
+from .search_gateway import SearchGateway, SearchRequest
+from .search_provider_registry import SearchProviderRegistry
+
+
+class SearchGatewayOrchestrator:
+    def __init__(self):
+        self.registry = SearchProviderRegistry()
+        self.audit = SearchAuditStore()
+        self.events = SearchEventBridge()
+        self.gateway = SearchGateway(self.registry.ordered())
+
+    def execute(self, payload: dict[str, Any]) -> dict[str, Any]:
+        request = SearchRequest(**payload)
+        self.events.search_started(request.to_dict())
+        response = self.gateway.search(request)
+        result = response.to_dict()
+        for attempt in result.get("attempts", []):
+            self.events.provider_attempted(result["request_id"], attempt)
+        if len(result.get("attempts", [])) > 1:
+            self.events.fallback_triggered(result["request_id"], "multiple providers attempted")
+        self.audit.append(result)
+        self.events.search_completed(result)
+        return result
+
+    def status(self) -> dict[str, Any]:
+        return {
+            "registry": self.registry.status(),
+            "audit": "enabled",
+            "events": "enabled",
+        }
+
+
+__all__ = ["SearchGatewayOrchestrator"]
@@ -0,0 +1,124 @@
+"""Search profile resolver for general-purpose AIOS Search Gateway.
+
+The Search Gateway is a shared platform capability. Groups and users may customize
+provider preference, query expansion hints, trusted sources and domain profiles on
+top of the shared base profile. Customization is intentionally layered and cannot
+override hard safety, authorization or audit rules.
+"""
+
+from __future__ import annotations
+
+from dataclasses import asdict, dataclass, field
+from typing import Any
+
+
+HARD_BLOCKED_BEHAVIORS = {
+    "bypass_login",
+    "bypass_captcha",
+    "scrape_private_student_grades",
+    "scrape_student_private_data",
+    "bulk_scrape_protected_platforms",
+    "collect_data_without_authorization",
+}
+
+
+@dataclass(frozen=True)
+class SearchProfile:
+    profile_id: str
+    scope: str
+    provider_order: list[str] = field(default_factory=lambda: [
+        "cache",
+        "official_site",
+        "general_web",
+        "vertical_source",
+        "browser_fetch",
+        "user_authorized_file",
+        "internal_archive",
+    ])
+    trusted_domains: list[str] = field(default_factory=list)
+    query_expansion_hints: list[str] = field(default_factory=list)
+    domain_profiles: list[str] = field(default_factory=list)
+    blocked_behaviors: set[str] = field(default_factory=lambda: set(HARD_BLOCKED_BEHAVIORS))
+    audit_required: bool = True
+    fallback_required: bool = True
+
+    def to_dict(self) -> dict[str, Any]:
+        data = asdict(self)
+        data["blocked_behaviors"] = sorted(self.blocked_behaviors)
+        return data
+
+
+class SearchProfileResolver:
+    """Resolves base -> group -> user -> session search preferences."""
+
+    def base_profile(self) -> SearchProfile:
+        return SearchProfile(profile_id="base", scope="global")
+
+    def resolve(
+        self,
+        group_profile: dict[str, Any] | None = None,
+        user_profile: dict[str, Any] | None = None,
+        session_profile: dict[str, Any] | None = None,
+    ) -> SearchProfile:
+        profile = self.base_profile().to_dict()
+        for overlay in (group_profile, user_profile, session_profile):
+            if overlay:
+                profile = self._merge(profile, overlay)
+        profile["blocked_behaviors"] = set(HARD_BLOCKED_BEHAVIORS)
+        profile["audit_required"] = True
+        profile["fallback_required"] = True
+        return SearchProfile(**profile)
+
+    def _merge(self, base: dict[str, Any], overlay: dict[str, Any]) -> dict[str, Any]:
+        merged = dict(base)
+        if overlay.get("provider_order"):
+            merged["provider_order"] = list(overlay["provider_order"])
+        for list_key in ("trusted_domains", "query_expansion_hints", "domain_profiles"):
+            merged[list_key] = self._dedupe(list(merged.get(list_key, [])) + list(overlay.get(list_key, [])))
+        merged["profile_id"] = overlay.get("profile_id", merged.get("profile_id", "custom"))
+        merged["scope"] = overlay.get("scope", merged.get("scope", "custom"))
+        return merged
+
+    def _dedupe(self, items: list[str]) -> list[str]:
+        return list(dict.fromkeys([item for item in items if item]))
+
+
+class DomainQueryProfileRegistry:
+    """Optional domain-specific expansions for the general Search Gateway."""
+
+    DEFAULT_PROFILES = {
+        "education_exam_data": [
+            "学校 年级 科目 试卷 答案 成绩统计",
+            "联考 月考 期中 期末 质量检测",
+        ],
+        "policy_regulation": [
+            "官方 通知 政策 解读 实施细则",
+            "site:gov.cn OR site:edu.cn",
+        ],
+        "software_library": [
+            "release notes changelog migration guide",
+            "official documentation GitHub issues",
+        ],
+        "product_research": [
+            "specs review comparison price availability",
+            "official page manual support",
+        ],
+        "company_research": [
+            "official announcement press release filings",
+            "leadership product financial report",
+        ],
+    }
+
+    def hints_for(self, domain_profiles: list[str]) -> list[str]:
+        hints: list[str] = []
+        for profile in domain_profiles:
+            hints.extend(self.DEFAULT_PROFILES.get(profile, []))
+        return list(dict.fromkeys(hints))
+
+
+__all__ = [
+    "HARD_BLOCKED_BEHAVIORS",
+    "SearchProfile",
+    "SearchProfileResolver",
+    "DomainQueryProfileRegistry",
+]
@@ -0,0 +1,121 @@
+"""Provider registry for the general AIOS Search Gateway.
+
+The registry makes provider availability explicit and auditable. Provider
+implementations are pluggable: production can register real public search,
+official-site, vertical-source, browser-fetch, authorized-file or archive
+providers without changing the Search Gateway core.
+"""
+
+from __future__ import annotations
+
+from dataclasses import asdict, dataclass, field
+from typing import Any
+
+from .search_gateway import SearchEvidence, SearchProviderAttempt, SearchRequest
+
+
+@dataclass(frozen=True)
+class SearchProviderConfig:
+    name: str
+    provider_type: str
+    enabled: bool = True
+    requires_secret: bool = False
+    supports_public_web: bool = True
+    supports_authorized_sources: bool = False
+    health: str = "unknown"
+    metadata: dict[str, Any] = field(default_factory=dict)
+
+    def to_dict(self) -> dict[str, Any]:
+        return asdict(self)
+
+
+class NullSearchProvider:
+    """Safe placeholder provider used until runtime wires a real adapter."""
+
+    def __init__(self, config: SearchProviderConfig):
+        self.config = config
+        self.name = config.name
+
+    def search(self, request: SearchRequest, queries: list[str]):
+        return [], SearchProviderAttempt(
+            provider=self.name,
+            status="not_configured" if self.config.enabled else "disabled",
+            result_count=0,
+            reason="provider adapter not wired in runtime",
+        )
+
+
+class StaticArchiveProvider:
+    """Small deterministic provider for tests and internal archived evidence."""
+
+    name = "internal_archive"
+
+    def __init__(self, records: list[dict[str, Any]] | None = None):
+        self.records = records or []
+
+    def search(self, request: SearchRequest, queries: list[str]):
+        evidence: list[SearchEvidence] = []
+        for record in self.records:
+            haystack = " ".join([record.get("title", ""), record.get("snippet", ""), record.get("tags", "")]).lower()
+            if any(q.lower() in haystack for q in queries):
+                evidence.append(
+                    SearchEvidence(
+                        title=record.get("title", "archive result"),
+                        url=record.get("url"),
+                        source=self.name,
+                        snippet=record.get("snippet", ""),
+                        evidence_type="internal_archive",
+                        confidence=record.get("confidence", "medium"),
+                    )
+                )
+        return evidence, SearchProviderAttempt(provider=self.name, status="ok", result_count=len(evidence))
+
+
+class SearchProviderRegistry:
+    DEFAULT_PROVIDER_CONFIGS = [
+        SearchProviderConfig("cache", "cache", supports_public_web=False),
+        SearchProviderConfig("official_site", "official_site"),
+        SearchProviderConfig("general_web", "general_web"),
+        SearchProviderConfig("vertical_source", "vertical_source"),
+        SearchProviderConfig("browser_fetch", "browser_fetch"),
+        SearchProviderConfig("user_authorized_file", "authorized_file", supports_public_web=False, supports_authorized_sources=True),
+        SearchProviderConfig("internal_archive", "archive", supports_public_web=False),
+    ]
+
+    def __init__(self):
+        self.configs: dict[str, SearchProviderConfig] = {}
+        self.providers: dict[str, Any] = {}
+        for config in self.DEFAULT_PROVIDER_CONFIGS:
+            self.register_config(config)
+
+    def register_config(self, config: SearchProviderConfig) -> None:
+        self.configs[config.name] = config
+        self.providers.setdefault(config.name, NullSearchProvider(config))
+
+    def register_provider(self, name: str, provider: Any, config: SearchProviderConfig | None = None) -> None:
+        if config:
+            self.configs[name] = config
+        elif name not in self.configs:
+            self.configs[name] = SearchProviderConfig(name=name, provider_type="custom")
+        self.providers[name] = provider
+
+    def ordered(self, order: list[str] | None = None) -> list[Any]:
+        order = order or list(self.configs)
+        result = [self.providers[name] for name in order if name in self.providers and self.configs[name].enabled]
+        result.extend([provider for name, provider in self.providers.items() if name not in order and self.configs[name].enabled])
+        return result
+
+    def status(self) -> dict[str, Any]:
+        return {
+            "providers": [self.configs[name].to_dict() for name in sorted(self.configs)],
+            "enabled": len([c for c in self.configs.values() if c.enabled]),
+            "configured": len([name for name, provider in self.providers.items() if not isinstance(provider, NullSearchProvider)]),
+        }
+
+
+__all__ = [
+    "SearchProviderConfig",
+    "NullSearchProvider",
+    "StaticArchiveProvider",
+    "SearchProviderRegistry",
+]
@@ -0,0 +1,265 @@
+<!DOCTYPE html>
+<html lang="en">
+<head>
+<meta charset="UTF-8">
+<meta name="viewport" content="width=device-width, initial-scale=1.0">
+<title>AI Evolution Dashboard</title>
+<style>
+* { margin: 0; padding: 0; box-sizing: border-box; }
+body { font-family: -apple-system, BlinkMacSystemFont, 'Segoe UI', Roboto, Oxygen, Ubuntu, Cantarell, sans-serif; background: #0d1117; color: #c9d1d9; height: 100vh; overflow: hidden; }
+.app { display: flex; height: 100vh; }
+.sidebar { width: 260px; background: #161b22; border-right: 1px solid #30363d; padding: 20px; display: flex; flex-direction: column; gap: 16px; }
+.sidebar h1 { font-size: 18px; color: #f0f6fc; border-bottom: 1px solid #30363d; padding-bottom: 12px; margin-bottom: 8px; }
+.sidebar-nav { list-style: none; display: flex; flex-direction: column; gap: 4px; }
+.sidebar-nav li { padding: 8px 12px; border-radius: 6px; cursor: pointer; font-size: 14px; color: #8b949e; }
+.sidebar-nav li:hover, .sidebar-nav li.active { background: #1f6feb22; color: #58a6ff; }
+.sidebar-nav li i { margin-right: 8px; }
+.main { flex: 1; padding: 24px; overflow-y: auto; }
+.main-header { display: flex; justify-content: space-between; align-items: center; margin-bottom: 24px; }
+.main-header h2 { font-size: 20px; }
+.status-badge { padding: 4px 12px; border-radius: 12px; font-size: 12px; font-weight: 600; text-transform: uppercase; }
+.status-badge.ok { background: #1b3a2d; color: #3fb950; border: 1px solid #3fb95044; }
+.status-badge.degraded { background: #3a2e1b; color: #d29922; border: 1px solid #d2992244; }
+.status-badge.fail { background: #3a1b1b; color: #f85149; border: 1px solid #f8514944; }
+.card { background: #161b22; border: 1px solid #30363d; border-radius: 8px; padding: 16px; margin-bottom: 16px; }
+.card-title { font-size: 14px; color: #8b949e; margin-bottom: 8px; }
+.card-value { font-size: 28px; font-weight: 700; color: #f0f6fc; }
+.row { display: flex; gap: 16px; }
+.col { flex: 1; }
+.grid { display: grid; grid-template-columns: repeat(auto-fill, minmax(300px, 1fr)); gap: 16px; }
+pre { white-space: pre-wrap; word-break: break-word; }
+.command-item { padding: 12px; margin-bottom: 8px; border-radius: 6px; background: #0d1117; border: 1px solid #21262d; }
+.command-item .meta { display: flex; gap: 8px; align-items: center; font-size: 12px; color: #8b949e; }
+.command-item .name { font-size: 14px; color: #58a6ff; font-weight: 600; }
+.command-item .state { font-size: 11px; padding: 2px 8px; border-radius: 10px; text-transform: uppercase; }
+.state-active { background: #1b3a2d; color: #3fb950; }
+.state-running { background: #3a2e1b; color: #d29922; }
+.state-completed { background: #1b3a2d; color: #3fb950; }
+.state-failed { background: #3a1b1b; color: #f85149; }
+.loading { text-align: center; padding: 40px; color: #8b949e; }
+.error { padding: 20px; background: #3a1b1b; border: 1px solid #f8514944; border-radius: 8px; color: #f85149; margin-bottom: 16px; }
+.version-item { padding: 8px 12px; background: #0d1117; border: 1px solid #21262d; border-radius: 6px; margin-bottom: 4px; display: flex; justify-content: space-between; }
+.version-item .tag { color: #58a6ff; font-weight: 600; }
+.version-item .name { color: #8b949e; font-size: 13px; }
+.agent-item { background: #0d1117; border: 1px solid #21262d; border-radius: 6px; margin-bottom: 8px; padding: 12px; }
+.agent-item .agent-title { font-size: 14px; font-weight: 600; color: #c9d1d9; margin-bottom: 4px; }
+.agent-item .agent-meta { font-size: 12px; color: #8b949e; }
+.agent-run-item { padding: 8px 12px; margin-bottom: 4px; border-radius: 4px; background: #0d1117; border: 1px solid #21262d; display: flex; justify-content: space-between; align-items: center; }
+.agent-run-item .run-id { color: #58a6ff; font-size: 13px; font-weight: 600; }
+.agent-run-item .source { color: #8b949e; font-size: 12px; }
+</style>
+</head>
+<body>
+<div id="app" class="app">
+  <div class="sidebar">
+    <h1>🕹️ AI Evolution Dashboard</h1>
+    <ul class="sidebar-nav">
+      <li class="active" onclick="showPanel('overview')"><i>📊</i> Overview</li>
+      <li onclick="showPanel('versions')"><i>🏷️</i> Versions</li>
+      <li onclick="showPanel('commands')"><i>⚡</i> Commands</li>
+      <li onclick="showPanel('agents')"><i>🤖</i> Agent Runs</li>
+      <li onclick="showPanel('arch-owner')"><i>🔐</i> Owner Gate</li>
+      <li onclick="showPanel('arch-approve')"><i>✓</i> Approve Changes</li>
+    </ul>
+  </div>
+  <div class="main">
+    <div id="panel-overview" class="panel">
+      <div class="main-header">
+        <h2>System Overview</h2>
+        <div id="header-status" class="status-badge">Loading...</div>
+      </div>
+      <div id="loading" class="loading">Loading...</div>
+      <div id="error" class="error" style="display:none;"></div>
+      <div id="content">
+        <div class="row">
+          <div class="col"><div class="card"><div class="card-title">Current Baseline</div><div id="base" class="card-value">-</div></div></div>
+          <div class="col"><div class="card"><div class="card-title">Git Branch</div><div id="branch" class="card-value">-</div></div></div>
+          <div class="col"><div class="card"><div class="card-title">Uptime</div><div id="uptime" class="card-value">-</div></div></div>
+        </div>
+        <div class="card" style="margin-top:16px">
+          <div class="card-title">Event Bus Status</div>
+          <pre id="bus-status">-</pre>
+        </div>
+      </div>
+    </div>
+    <div id="panel-versions" class="panel" style="display:none;">
+      <div class="main-header"><h2>Version Timeline</h2></div>
+      <div id="versions-list">Loading...</div>
+    </div>
+    <div id="panel-commands" class="panel" style="display:none;">
+      <div class="main-header"><h2>Command Lookup</h2></div>
+      <div id="commands-list">Loading...</div>
+    </div>
+    <div id="panel-agents" class="panel" style="display:none;">
+      <div class="main-header"><h2>Recent Agent Executions</h2></div>
+    <div id="panel-arch-owner" class="panel" style="display:none;">
+      <div class="main-header"><h2>🔐 Architecture Owner Gate</h2></div>
+      <div id="arch-owner-content">Loading...</div>
+    </div>
+    <div id="panel-arch-approve" class="panel" style="display:none;">
+      <div class="main-header"><h2>✓ Approve Architecture Changes</h2></div>
+      <div id="arch-approve-content">Loading...</div>
+    </div>
+      <div id="agents-list">Loading...</div>
+    </div>
+  </div>
+</div>
+<script>
+async function fetchData() {
+  const r = await fetch('/admin/ai-evolution/api/data');
+  const data = await r.json();
+  document.getElementById('loading').style.display = 'none';
+  document.getElementById('error').style.display = 'none';
+  document.getElementById('content').style.display = '';
+
+  const hs = document.getElementById('header-status');
+  hs.className = 'status-badge ' + (data.healthy ? 'ok' : 'fail');
+  hs.textContent = data.healthy ? '▪ operational' : '⚠ degraded';
+
+  document.getElementById('base').textContent = data.baseline || '-';
+  document.getElementById('branch').textContent = data.main_branch || 'main';
+  document.getElementById('uptime').textContent = data.uptime || '-';
+  document.getElementById('bus-status').textContent = data.bus ? JSON.stringify(data.bus, null, 2) : 'no event bus data';
+}
+
+async function loadVersions() {
+  const el = document.getElementById('versions-list');
+  el.innerHTML = '<div class="loading">Loading versions...</div>';
+  try {
+    const r = await fetch('/admin/ai-evolution/api/versions');
+    const data = await r.json();
+    if (data.error) { el.innerHTML = '<div class="error">' + data.error + '</div>'; return; }
+    let html = '<div class="grid">';
+    for (const v of data.versions || []) {
+      html += '<div class="card"><div class="card-title">' + v.tag + '</div><div class="card-value">' + (v.version || v.tag) + '</div></div>';
+    }
+    html += '</div>';
+    el.innerHTML = html;
+  } catch(e) {
+    el.innerHTML = '<div class="error">Failed to load versions: ' + e.message + '</div>';
+  }
+}
+
+async function loadCommands() {
+  const el = document.getElementById('commands-list');
+  el.innerHTML = '<div class="loading">Loading commands...</div>';
+  try {
+    const r = await fetch('/admin/ai-evolution/api/data');
+    const data = await r.json();
+    let cmds = data.commands || [];
+    let html = '<div style="display:flex;gap:24px;margin-bottom:16px;">';
+    html += '<div><div class="card"><div class="card-title">Total</div><div class="card-value">' + cmds.length + '</div></div></div>';
+    html += '<div><div class="card"><div class="card-title">By Category</div>';
+    const cats = {};
+    for (const c of cmds) { const k = c.category || 'uncategorized'; cats[k] = (cats[k] || 0) + 1; }
+    for (const [k, v] of Object.entries(cats)) html += '<div style="font-size:14px;color:#8b949e;margin:4px 0;">' + k + ': ' + v + '</div>';
+    html += '</div></div></div>';
+    for (const cmd of cmds.slice().reverse()) {
+      html += '<div class="command-item"><div class="meta">' + cmd.category + ' | ' + cmd.filename + '</div><div class="name">' + cmd.title + '</div><div style="font-size:13px;color:#c9d1d9;margin-top:4px;">' + cmd.summary + '</div></div>';
+    }
+    el.innerHTML = html;
+  } catch(e) {
+    el.innerHTML = '<div class="error">Failed to load commands: ' + e.message + '</div>';
+  }
+}
+
+async function loadAgents() {
+  const el = document.getElementById('agents-list');
+  el.innerHTML = '<div class="loading">Loading agent runs...</div>';
+  try {
+    const r = await fetch('/admin/ai-evolution/api/agents');
+    const data = await r.json();
+    if (data.error) { el.innerHTML = '<div class="error">' + data.error + '</div>'; return; }
+    let html = '';
+    for (const agent of data.agents || []) {
+      html += '<div class="agent-item"><div class="agent-title">' + agent.name + ' <span class="state state-' + (agent.status === 'running' ? 'running' : (agent.status === 'ok' ? 'completed' : 'failed')) + '">' + agent.status + '</span></div><div class="agent-meta">' + (agent.runs || 0) + ' runs | last: ' + (agent.last_run || '-') + '</div></div>';
+    }
+    el.innerHTML = html || '<div style="padding:24px;color:#8b949e;">No agent runs recorded yet.</div>';
+  } catch(e) {
+    el.innerHTML = '<div class="error">Failed to load agents: ' + e.message + '</div>';
+  }
+}
+
+function showPanel(name) {
+
+async function loadOwnerGate() {
+  const el = document.getElementById('arch-owner-content');
+  el.innerHTML = '<div class="loading">Loading owner gate data...</div>';
+  try {
+    const r = await fetch('/admin/ai-evolution/api/owner-gate');
+    const data = await r.json();
+    let html = '<div class="row">';
+    html += '<div class="col"><div class="card"><div class="card-title">Pending Requests</div><div class="card-value">' + data.pending_count + '</div></div></div>';
+    html += '<div class="col"><div class="card"><div class="card-title">Approved</div><div class="card-value">' + data.approved_count + '</div></div></div>';
+    html += '<div class="col"><div class="card"><div class="card-title">Rejected</div><div class="card-value">' + data.rejected_count + '</div></div></div>';
+    html += '</div>';
+    html += '<div class="card"><div class="card-title">Pending Approval Requests</div>';
+    if (data.pending_requests && data.pending_requests.length) {
+      for (const req of data.pending_requests) {
+        html += '<div class="command-item">';
+        html += '<div class="name">' + (req.title || req.change_id) + '</div>';
+        html += '<div class="meta">ID: ' + req.change_id + ' | Risk: ' + req.risk + '</div>';
+        html += '<div style="font-size:13px;color:#c9d1d9;margin-top:4px;">' + (req.reason || '') + '</div>';
+        html += '</div>';
+      }
+    } else {
+      html += '<div style="color:#8b949e;padding:12px;">No pending requests.</div>';
+    }
+    html += '</div>';
+    el.innerHTML = html;
+  } catch(e) {
+    el.innerHTML = '<div class="error">Failed to load: ' + e.message + '</div>';
+  }
+}
+
+async function loadApprove() {
+  const el = document.getElementById('arch-approve-content');
+  el.innerHTML = '<div class="loading">Loading pending approvals...</div>';
+  try {
+    const r = await fetch('/admin/ai-evolution/api/owner-gate');
+    const data = await r.json();
+    let html = '';
+    if (data.pending_requests && data.pending_requests.length) {
+      html += '<div class="card"><div class="card-title">Pending changes requiring your approval</div>';
+      for (const req of data.pending_requests) {
+        html += '<div class="command-item">';
+        html += '<div class="name">' + (req.title || req.change_id) + '</div>';
+        html += '<div class="meta">Risk: <strong>' + req.risk + '</strong></div>';
+        html += '<div style="font-size:13px;color:#c9d1d9;margin:8px 0;">' + (req.reason || '') + '</div>';
+        html += '<div style="display:flex;gap:8px;">';
+        html += '<button onclick="ownerAction.classList.add('\'approved\');this.disabled=true" style="padding:6px 16px;background:#238636;color:#fff;border:none;border-radius:6px;cursor:pointer;font-size:13px;">✓ Approve</button>';
+        html += '</div></div>';
+      }
+      html += '</div>';
+    } else {
+      html += '<div class="card"><div style="padding:24px;text-align:center;color:#8b949e;">✅ All changes processed. No pending approval requests.</div></div>';
+    }
+    el.innerHTML = html;
+  } catch(e) {
+    el.innerHTML = '<div class="error">Failed to load: ' + e.message + '</div>';
+  }
+}
+
+  document.querySelectorAll('.panel').forEach(p => p.style.display = 'none');
+  document.getElementById('panel-' + name).style.display = '';
+  document.querySelectorAll('.sidebar-nav li').forEach(li => li.classList.remove('active'));
+  document.querySelector('.sidebar-nav li:nth-child(' + ({overview:1,versions:2,commands:3,agents:4}[name] || 1) + ')').classList.add('active');
+  if (name === 'overview') {
+    document.getElementById('panel-overview').innerHTML = '<div class="main-header"><h2>System Overview</h2><div id="header-status" class="status-badge">Loading...</div></div><div id="loading" class="loading">Loading...</div><div id="error" class="error" style="display:none;"></div><div id="content"><div class="row"><div class="col"><div class="card"><div class="card-title">Current Baseline</div><div id="base" class="card-value">-</div></div></div><div class="col"><div class="card"><div class="card-title">Git Branch</div><div id="branch" class="card-value">-</div></div></div><div class="col"><div class="card"><div class="card-title">Uptime</div><div id="uptime" class="card-value">-</div></div></div></div><div class="card" style="margin-top:16px"><div class="card-title">Event Bus Status</div><pre id="bus-status">-</pre></div></div>';
+    fetchData();
+  }
+  const loaders = {
+    'arch-owner': loadOwnerGate,
+    'arch-approve': loadApprove,
+    versions: loadVersions,
+    commands: loadCommands,
+    agents: loadAgents,
+  };
+  const fn = loaders[name];
+  if (fn) fn();
+}
+
+fetchData();
+</script>
+</body>
+</html>
@@ -0,0 +1,41 @@
+"""Task Center action API model for Web UI buttons."""
+
+from typing import Any
+
+try:
+    from fastapi import APIRouter
+except Exception:
+    APIRouter = None
+
+from .task_lifecycle import TaskLifecycle
+
+
+lifecycle = TaskLifecycle()
+
+
+def get_task_actions(task: dict[str, Any], role: str) -> dict[str, Any]:
+    return {
+        "task_id": task.get("task_id"),
+        "actions": lifecycle.available_actions(task, role),
+    }
+
+
+def execute_task_action(task: dict[str, Any], action: str, role: str, reason: str | None = None):
+    return lifecycle.transition(task, action, role, reason)
+
+
+def build_router():
+    if APIRouter is None:
+        return None
+    router = APIRouter(prefix="/admin/ai-evolution/tasks", tags=["task-actions"])
+
+    @router.post("/{task_id}/actions/{action}", operation_id="executeTaskAction")
+    def action(task_id: str, action: str, payload: dict[str, Any]):
+        return execute_task_action(payload.get("task", {}), action, payload.get("role", "user"), payload.get("reason"))
+
+    return router
+
+
+router = build_router()
+
+__all__ = ["router", "get_task_actions", "execute_task_action"]
@@ -0,0 +1,115 @@
+"""AIOS Task Center domain model.
+
+The Task Center gives owners and authorized users a single Web UI page showing:
+who assigned a task, who is executing it, progress, elapsed time, remaining
+work, and whether completion notifications were sent.
+"""
+
+from __future__ import annotations
+
+from dataclasses import asdict, dataclass, field
+from datetime import datetime, timezone
+from typing import Any
+
+
+def utc_now() -> str:
+    return datetime.now(timezone.utc).isoformat()
+
+
+def parse_time(value: str | None) -> datetime | None:
+    if not value:
+        return None
+    return datetime.fromisoformat(value.replace("Z", "+00:00"))
+
+
+@dataclass(frozen=True)
+class TaskNotificationStatus:
+    required: bool = True
+    sent: bool = False
+    channel: str | None = None
+    sent_at: str | None = None
+    recipient_scope: str | None = None
+    last_error: str | None = None
+
+    def to_dict(self) -> dict[str, Any]:
+        return asdict(self)
+
+
+@dataclass(frozen=True)
+class TaskProgress:
+    percent: int = 0
+    current_step: str = "pending"
+    completed_steps: list[str] = field(default_factory=list)
+    remaining_steps: list[str] = field(default_factory=list)
+    blocked_reason: str | None = None
+
+    def to_dict(self) -> dict[str, Any]:
+        return asdict(self)
+
+
+@dataclass(frozen=True)
+class AIOSTaskRecord:
+    task_id: str
+    title: str
+    assigned_by: str
+    executor: str
+    source: str
+    status: str = "pending"
+    priority: str = "normal"
+    created_at: str = field(default_factory=utc_now)
+    started_at: str | None = None
+    updated_at: str = field(default_factory=utc_now)
+    completed_at: str | None = None
+    progress: TaskProgress = field(default_factory=TaskProgress)
+    notification: TaskNotificationStatus = field(default_factory=TaskNotificationStatus)
+    owner_gate_required: bool = False
+    owner_gate_status: str | None = None
+    evidence: list[str] = field(default_factory=list)
+
+    def elapsed_seconds(self) -> int:
+        start = parse_time(self.started_at) or parse_time(self.created_at)
+        end = parse_time(self.completed_at) or datetime.now(timezone.utc)
+        return int((end - start).total_seconds()) if start else 0
+
+    def to_dict(self) -> dict[str, Any]:
+        data = asdict(self)
+        data["elapsed_seconds"] = self.elapsed_seconds()
+        data["progress"] = self.progress.to_dict()
+        data["notification"] = self.notification.to_dict()
+        return data
+
+
+class TaskCenter:
+    def summarize(self, tasks: list[AIOSTaskRecord] | list[dict[str, Any]] | None = None) -> dict[str, Any]:
+        raw_tasks = tasks or []
+        task_dicts = [t.to_dict() if hasattr(t, "to_dict") else dict(t) for t in raw_tasks]
+        active = [t for t in task_dicts if t.get("status") in {"running", "pending", "blocked"}]
+        blocked = [t for t in task_dicts if t.get("status") == "blocked"]
+        completed = [t for t in task_dicts if t.get("status") == "completed"]
+        notification_pending = [
+            t for t in completed
+            if t.get("notification", {}).get("required") and not t.get("notification", {}).get("sent")
+        ]
+        return {
+            "module": "task-center",
+            "route": "/admin/ai-evolution/tasks",
+            "title": "AIOS 任务中心",
+            "status": "attention_required" if blocked or notification_pending else "healthy",
+            "generated_at": utc_now(),
+            "summary": {
+                "total": len(task_dicts),
+                "active": len(active),
+                "blocked": len(blocked),
+                "completed": len(completed),
+                "notification_pending": len(notification_pending),
+            },
+            "tasks": sorted(task_dicts, key=lambda t: t.get("updated_at", ""), reverse=True),
+        }
+
+
+__all__ = [
+    "TaskNotificationStatus",
+    "TaskProgress",
+    "AIOSTaskRecord",
+    "TaskCenter",
+]
@@ -0,0 +1,34 @@
+"""Task Center API."""
+
+from typing import Any
+
+try:
+    from fastapi import APIRouter
+except Exception:
+    APIRouter = None
+
+from .task_center import TaskCenter
+
+
+task_center = TaskCenter()
+
+
+def get_task_overview(tasks: list[dict[str, Any]] | None = None):
+    return task_center.summarize(tasks)
+
+
+def build_router():
+    if APIRouter is None:
+        return None
+    router = APIRouter(prefix="/admin/ai-evolution/tasks", tags=["task-center"])
+
+    @router.get("/overview", operation_id="getTaskCenterOverview")
+    def overview():
+        return get_task_overview()
+
+    return router
+
+
+router = build_router()
+
+__all__ = ["router", "get_task_overview"]
@@ -0,0 +1,81 @@
+"""Task Center visualization model."""
+
+from __future__ import annotations
+
+from typing import Any
+
+
+class TaskDashboardView:
+    def build(self, tasks: list[dict[str, Any]] | None = None) -> dict[str, Any]:
+        tasks = tasks or []
+        return {
+            "module": "task-center",
+            "route": "/admin/ai-evolution/tasks",
+            "title": "AIOS Task Center",
+            "views": {
+                "kanban": self._kanban(tasks),
+                "timeline": self._timeline(tasks),
+                "flow": self._flow(tasks),
+                "statistics": self._statistics(tasks),
+            },
+            "tasks": [self._task_card(task) for task in tasks],
+        }
+
+    def _task_card(self, task: dict[str, Any]) -> dict[str, Any]:
+        return {
+            "task_id": task.get("task_id"),
+            "title": task.get("title"),
+            "assigned_by": task.get("assigned_by"),
+            "executor": task.get("executor"),
+            "status": task.get("status"),
+            "progress": task.get("progress", {}),
+            "elapsed_seconds": task.get("elapsed_seconds", 0),
+            "remaining_steps": task.get("progress", {}).get("remaining_steps", []),
+            "notification": task.get("notification", {}),
+            "actions": [
+                "start",
+                "update_progress",
+                "block",
+                "request_approval",
+                "approve",
+                "reject",
+                "retry",
+                "complete",
+                "notify",
+                "close",
+            ],
+        }
+
+    def _kanban(self, tasks):
+        columns = {"pending": [], "running": [], "blocked": [], "waiting_approval": [], "completed": []}
+        for task in tasks:
+            columns.setdefault(task.get("status", "pending"), []).append(task.get("task_id"))
+        return columns
+
+    def _timeline(self, tasks):
+        return [
+            {
+                "task_id": task.get("task_id"),
+                "updated_at": task.get("updated_at"),
+                "status": task.get("status"),
+            }
+            for task in tasks
+        ]
+
+    def _flow(self, tasks):
+        return {
+            "nodes": ["created", "assigned", "running", "blocked", "approval", "completed", "closed"],
+            "task_count": len(tasks),
+        }
+
+    def _statistics(self, tasks):
+        return {
+            "total": len(tasks),
+            "running": len([t for t in tasks if t.get("status") == "running"]),
+            "blocked": len([t for t in tasks if t.get("status") == "blocked"]),
+            "waiting_approval": len([t for t in tasks if t.get("status") == "waiting_approval"]),
+            "completed": len([t for t in tasks if t.get("status") == "completed"]),
+        }
+
+
+__all__ = ["TaskDashboardView"]
@@ -0,0 +1,129 @@
+"""Task Center event bridge.
+
+Converts task lifecycle actions and agent execution events into normalized AIOS
+Task Center events. Uses duck-typed Agent Event Bus publisher when available and
+an in-memory fallback for tests/offline runs.
+"""
+
+from __future__ import annotations
+
+from dataclasses import asdict, dataclass, field
+from datetime import datetime, timezone
+from typing import Any, Protocol
+from uuid import uuid4
+
+
+def utc_now() -> str:
+    return datetime.now(timezone.utc).isoformat()
+
+
+class TaskEventPublisher(Protocol):
+    def publish(
+        self,
+        kind: str,
+        source: str,
+        target: str,
+        summary: str,
+        payload: dict[str, Any],
+        correlation_id: str | None = None,
+    ) -> dict[str, Any]:
+        ...
+
+
+@dataclass(frozen=True)
+class TaskEvent:
+    id: str
+    kind: str
+    source: str
+    target: str
+    summary: str
+    payload: dict[str, Any]
+    correlation_id: str | None = None
+    created_at: str = field(default_factory=utc_now)
+
+    def to_dict(self) -> dict[str, Any]:
+        return asdict(self)
+
+
+class InMemoryTaskEventBus:
+    def __init__(self):
+        self.events: list[dict[str, Any]] = []
+
+    def publish(
+        self,
+        kind: str,
+        source: str,
+        target: str,
+        summary: str,
+        payload: dict[str, Any],
+        correlation_id: str | None = None,
+    ) -> dict[str, Any]:
+        event = TaskEvent(
+            id=f"task_evt_{uuid4().hex[:12]}",
+            kind=kind,
+            source=source,
+            target=target,
+            summary=summary,
+            payload=payload,
+            correlation_id=correlation_id,
+        ).to_dict()
+        self.events.append(event)
+        return event
+
+
+class TaskEventBridge:
+    def __init__(self, bus: TaskEventPublisher | None = None):
+        self.bus = bus or InMemoryTaskEventBus()
+
+    def task_created(self, task: dict[str, Any]) -> dict[str, Any]:
+        return self.bus.publish(
+            kind="task_created",
+            source=task.get("assigned_by", "unknown"),
+            target=task.get("executor", "task-center"),
+            summary=task.get("title", "task created"),
+            payload=task,
+            correlation_id=task.get("task_id"),
+        )
+
+    def task_updated(self, task: dict[str, Any], action: str) -> dict[str, Any]:
+        return self.bus.publish(
+            kind="task_updated",
+            source=task.get("executor", "task-center"),
+            target="task-center",
+            summary=f"task {action}: {task.get('status')}",
+            payload={"action": action, "task": task},
+            correlation_id=task.get("task_id"),
+        )
+
+    def task_blocked(self, task: dict[str, Any], reason: str) -> dict[str, Any]:
+        return self.bus.publish(
+            kind="task_blocked",
+            source=task.get("executor", "unknown"),
+            target="owner-gate",
+            summary=reason,
+            payload={"task": task, "reason": reason},
+            correlation_id=task.get("task_id"),
+        )
+
+    def approval_requested(self, task: dict[str, Any], reason: str) -> dict[str, Any]:
+        return self.bus.publish(
+            kind="task_approval_requested",
+            source=task.get("executor", "unknown"),
+            target="owner-gate",
+            summary=reason,
+            payload={"task": task, "reason": reason},
+            correlation_id=task.get("task_id"),
+        )
+
+    def notification_sent(self, task: dict[str, Any], channel: str) -> dict[str, Any]:
+        return self.bus.publish(
+            kind="task_notification_sent",
+            source="notification-agent",
+            target=task.get("assigned_by", "owner"),
+            summary=f"notification sent via {channel}",
+            payload={"task": task, "channel": channel},
+            correlation_id=task.get("task_id"),
+        )
+
+
+__all__ = ["TaskEvent", "InMemoryTaskEventBus", "TaskEventBridge"]
@@ -0,0 +1,147 @@
+"""AIOS Task Lifecycle.
+
+Defines a governed task state machine for task creation, assignment, execution,
+blocking, approval, retry, completion, notification and closure.
+"""
+
+from __future__ import annotations
+
+from dataclasses import asdict, dataclass, field
+from datetime import datetime, timezone
+from typing import Any
+
+
+def utc_now() -> str:
+    return datetime.now(timezone.utc).isoformat()
+
+
+TERMINAL_STATES = {"completed", "cancelled", "failed", "closed"}
+
+
+@dataclass(frozen=True)
+class TaskTransition:
+    action: str
+    from_states: set[str]
+    to_state: str
+    required_role: str = "agent"
+    requires_reason: bool = False
+    notify: bool = False
+
+    def to_dict(self) -> dict[str, Any]:
+        data = asdict(self)
+        data["from_states"] = sorted(self.from_states)
+        return data
+
+
+class TaskLifecycle:
+    """Central task state machine."""
+
+    TRANSITIONS = {
+        "assign": TaskTransition("assign", {"new", "pending"}, "assigned", "dispatcher"),
+        "start": TaskTransition("start", {"assigned", "pending"}, "running", "executor"),
+        "update_progress": TaskTransition("update_progress", {"running"}, "running", "executor"),
+        "block": TaskTransition("block", {"running", "assigned", "pending"}, "blocked", "executor", requires_reason=True, notify=True),
+        "request_approval": TaskTransition("request_approval", {"running", "blocked"}, "waiting_approval", "executor", requires_reason=True, notify=True),
+        "approve": TaskTransition("approve", {"waiting_approval"}, "approved", "owner", notify=True),
+        "reject": TaskTransition("reject", {"waiting_approval"}, "rejected", "owner", requires_reason=True, notify=True),
+        "resume": TaskTransition("resume", {"blocked", "approved"}, "running", "executor"),
+        "retry": TaskTransition("retry", {"failed", "blocked"}, "running", "executor", notify=True),
+        "complete": TaskTransition("complete", {"running", "approved"}, "completed", "executor", notify=True),
+        "notify": TaskTransition("notify", {"completed"}, "completed", "notifier"),
+        "close": TaskTransition("close", {"completed", "cancelled", "failed"}, "closed", "owner"),
+        "cancel": TaskTransition("cancel", {"new", "pending", "assigned", "running", "blocked", "waiting_approval"}, "cancelled", "owner", requires_reason=True, notify=True),
+        "fail": TaskTransition("fail", {"running", "assigned"}, "failed", "executor", requires_reason=True, notify=True),
+    }
+
+    ROLE_ORDER = {
+        "user": 1,
+        "agent": 2,
+        "executor": 3,
+        "dispatcher": 4,
+        "notifier": 4,
+        "group_admin": 5,
+        "owner": 9,
+    }
+
+    def available_actions(self, task: dict[str, Any], role: str) -> list[dict[str, Any]]:
+        state = task.get("status", "new")
+        actions = []
+        for name, transition in self.TRANSITIONS.items():
+            allowed_state = state in transition.from_states
+            allowed_role = self._role_allows(role, transition.required_role)
+            actions.append({
+                "action": name,
+                "label": self._label(name),
+                "enabled": allowed_state and allowed_role,
+                "from_state": state,
+                "to_state": transition.to_state,
+                "required_role": transition.required_role,
+                "requires_reason": transition.requires_reason,
+                "notify": transition.notify,
+                "disabled_reason": None if allowed_state and allowed_role else self._disabled_reason(allowed_state, allowed_role, transition),
+            })
+        return actions
+
+    def transition(self, task: dict[str, Any], action: str, actor_role: str, reason: str | None = None) -> dict[str, Any]:
+        if action not in self.TRANSITIONS:
+            raise ValueError(f"unknown task action: {action}")
+        transition = self.TRANSITIONS[action]
+        state = task.get("status", "new")
+        if state not in transition.from_states:
+            raise ValueError(f"action {action} not allowed from state {state}")
+        if not self._role_allows(actor_role, transition.required_role):
+            raise PermissionError(f"role {actor_role} cannot execute {action}; requires {transition.required_role}")
+        if transition.requires_reason and not reason:
+            raise ValueError(f"action {action} requires reason")
+        updated = dict(task)
+        updated["status"] = transition.to_state
+        updated["updated_at"] = utc_now()
+        history = list(updated.get("history", []))
+        history.append({
+            "action": action,
+            "from": state,
+            "to": transition.to_state,
+            "actor_role": actor_role,
+            "reason": reason,
+            "ts": utc_now(),
+        })
+        updated["history"] = history
+        if action == "complete":
+            updated["completed_at"] = utc_now()
+            notification = dict(updated.get("notification", {}))
+            notification.setdefault("required", True)
+            notification.setdefault("sent", False)
+            updated["notification"] = notification
+        return updated
+
+    def _role_allows(self, role: str, required: str) -> bool:
+        return self.ROLE_ORDER.get(role, 0) >= self.ROLE_ORDER.get(required, 0)
+
+    def _label(self, action: str) -> str:
+        labels = {
+            "assign": "分配任务",
+            "start": "开始执行",
+            "update_progress": "更新进度",
+            "block": "标记阻塞",
+            "request_approval": "请求审批",
+            "approve": "批准",
+            "reject": "拒绝",
+            "resume": "继续执行",
+            "retry": "重试",
+            "complete": "标记完成",
+            "notify": "发送通知",
+            "close": "关闭任务",
+            "cancel": "取消任务",
+            "fail": "标记失败",
+        }
+        return labels.get(action, action)
+
+    def _disabled_reason(self, allowed_state: bool, allowed_role: bool, transition: TaskTransition) -> str:
+        if not allowed_state:
+            return "当前状态不允许该操作"
+        if not allowed_role:
+            return f"需要角色: {transition.required_role}"
+        return "不可用"
+
+
+__all__ = ["TaskTransition", "TaskLifecycle", "TERMINAL_STATES"]
@@ -0,0 +1,50 @@
+"""Task Center orchestrator.
+
+Coordinates task ingestion, persistence, lifecycle updates, events and statistics.
+"""
+
+from __future__ import annotations
+
+from typing import Any
+
+from .task_event_bridge import TaskEventBridge
+from .task_source_ingestor import TaskSourceIngestor
+from .task_statistics import TaskStatistics
+from .task_store import TaskStore
+
+
+class TaskOrchestrator:
+    def __init__(self):
+        self.ingestor = TaskSourceIngestor()
+        self.store = TaskStore()
+        self.events = TaskEventBridge()
+        self.statistics = TaskStatistics()
+
+    def create_from_command(self, path: str, bucket: str = "active") -> dict[str, Any]:
+        task = self.ingestor.command_file_to_task(path, bucket)
+        self.store.append(task)
+        self.events.task_created(task)
+        return task
+
+    def create_from_agent_execution(self, record: dict[str, Any]) -> dict[str, Any]:
+        task = self.ingestor.agent_execution_to_task(record)
+        self.store.append(task)
+        self.events.task_created(task)
+        return task
+
+    def sync_event(self, event: dict[str, Any]) -> dict[str, Any] | None:
+        task = self.ingestor.event_to_task(event)
+        if task:
+            self.store.append(task)
+            self.events.task_updated(task, event.get("kind", "event"))
+        return task
+
+    def dashboard(self) -> dict[str, Any]:
+        tasks = self.store.latest()
+        return {
+            "tasks": tasks,
+            "statistics": self.statistics.summarize(tasks),
+        }
+
+
+__all__ = ["TaskOrchestrator"]
@@ -0,0 +1,159 @@
+"""Task source ingestion for AIOS Task Center.
+
+Converts Command Watcher files, OpenClaw/Hermes/Codex execution records and
+Generic Agent Event Bus records into normalized Task Center task dictionaries.
+"""
+
+from __future__ import annotations
+
+from dataclasses import asdict, dataclass, field
+from datetime import datetime, timezone
+from pathlib import Path
+from typing import Any
+
+
+def utc_now() -> str:
+    return datetime.now(timezone.utc).isoformat()
+
+
+@dataclass(frozen=True)
+class TaskSourceRecord:
+    source_type: str
+    source_id: str
+    title: str
+    assigned_by: str
+    executor: str
+    status: str
+    priority: str = "normal"
+    created_at: str = field(default_factory=utc_now)
+    updated_at: str = field(default_factory=utc_now)
+    evidence: list[str] = field(default_factory=list)
+    metadata: dict[str, Any] = field(default_factory=dict)
+
+    def to_task(self) -> dict[str, Any]:
+        return {
+            "task_id": f"task-{self.source_type}-{self.source_id}",
+            "title": self.title,
+            "assigned_by": self.assigned_by,
+            "executor": self.executor,
+            "source": self.source_type,
+            "status": self.status,
+            "priority": self.priority,
+            "created_at": self.created_at,
+            "updated_at": self.updated_at,
+            "progress": {
+                "percent": self._progress_percent(),
+                "current_step": self.status,
+                "completed_steps": [],
+                "remaining_steps": [],
+                "blocked_reason": self.metadata.get("blocked_reason"),
+            },
+            "notification": {
+                "required": True,
+                "sent": bool(self.metadata.get("notification_sent")),
+                "channel": self.metadata.get("notification_channel"),
+                "sent_at": self.metadata.get("notification_sent_at"),
+            },
+            "evidence": self.evidence,
+            "metadata": self.metadata,
+        }
+
+    def _progress_percent(self) -> int:
+        mapping = {
+            "new": 0,
+            "pending": 5,
+            "assigned": 15,
+            "running": 50,
+            "blocked": 50,
+            "waiting_approval": 60,
+            "approved": 70,
+            "completed": 100,
+            "closed": 100,
+            "failed": 100,
+            "cancelled": 100,
+        }
+        return mapping.get(self.status, 0)
+
+
+class TaskSourceIngestor:
+    COMMAND_STATUS_MAP = {
+        "active": "pending",
+        "running": "running",
+        "completed": "completed",
+        "failed": "failed",
+        "archive": "closed",
+    }
+
+    def command_file_to_task(self, path: str, bucket: str = "active") -> dict[str, Any]:
+        p = Path(path)
+        title = p.stem
+        return TaskSourceRecord(
+            source_type="command_watcher",
+            source_id=title,
+            title=title,
+            assigned_by="command-watcher",
+            executor=self._infer_executor(title),
+            status=self.COMMAND_STATUS_MAP.get(bucket, "pending"),
+            priority="high" if "P0" in title or "fix" in title.lower() else "normal",
+            evidence=[str(p)],
+            metadata={"bucket": bucket},
+        ).to_task()
+
+    def agent_execution_to_task(self, record: dict[str, Any]) -> dict[str, Any]:
+        source_id = record.get("execution_id", record.get("id", "unknown"))
+        status = record.get("status", "running")
+        return TaskSourceRecord(
+            source_type="agent_execution",
+            source_id=source_id,
+            title=record.get("title", f"Agent execution {source_id}"),
+            assigned_by=record.get("assigned_by", record.get("entry_agent", "agent-system")),
+            executor=record.get("closed_by", record.get("entry_agent", "unknown-agent")),
+            status=status,
+            priority=record.get("priority", "normal"),
+            created_at=record.get("started_at", utc_now()),
+            updated_at=record.get("ended_at", utc_now()),
+            evidence=[record.get("execution_id", source_id)],
+            metadata=record,
+        ).to_task()
+
+    def event_to_task(self, event: dict[str, Any]) -> dict[str, Any] | None:
+        kind = event.get("kind")
+        payload = event.get("payload", {})
+        if kind not in {"task_created", "task_updated", "agent_started", "agent_completed", "agent_failed", "search_started", "search_completed"}:
+            return None
+        source_id = event.get("correlation_id") or event.get("id", "event")
+        status = self._status_from_event(kind, payload)
+        return TaskSourceRecord(
+            source_type="event_bus",
+            source_id=source_id,
+            title=payload.get("title") or event.get("summary", kind),
+            assigned_by=event.get("source", "event-bus"),
+            executor=event.get("target", "task-center"),
+            status=status,
+            created_at=event.get("created_at", utc_now()),
+            updated_at=event.get("created_at", utc_now()),
+            evidence=[event.get("id", source_id)],
+            metadata={"event_kind": kind, "payload": payload},
+        ).to_task()
+
+    def _status_from_event(self, kind: str, payload: dict[str, Any]) -> str:
+        if kind.endswith("completed"):
+            return "completed"
+        if kind.endswith("failed"):
+            return "failed"
+        return payload.get("status", "running")
+
+    def _infer_executor(self, title: str) -> str:
+        lower = title.lower()
+        if "openclaw" in lower:
+            return "openclaw"
+        if "hermes" in lower:
+            return "hermes"
+        if "codex" in lower:
+            return "codex"
+        if "search" in lower:
+            return "openclaw"
+        return "openclaw"
+
+
+__all__ = ["TaskSourceRecord", "TaskSourceIngestor"]
@@ -0,0 +1,33 @@
+"""Task Center statistics."""
+
+from typing import Any
+
+
+class TaskStatistics:
+    def summarize(self, tasks: list[dict[str, Any]] | None = None) -> dict[str, Any]:
+        tasks = tasks or []
+        completed = [t for t in tasks if t.get("status") == "completed"]
+        blocked = [t for t in tasks if t.get("status") == "blocked"]
+        waiting = [t for t in tasks if t.get("status") == "waiting_approval"]
+        elapsed = [t.get("elapsed_seconds", 0) for t in tasks if t.get("elapsed_seconds") is not None]
+        return {
+            "total": len(tasks),
+            "completed": len(completed),
+            "blocked": len(blocked),
+            "waiting_approval": len(waiting),
+            "completion_rate": round(len(completed) / len(tasks) * 100, 2) if tasks else 0,
+            "average_elapsed_seconds": round(sum(elapsed) / len(elapsed), 2) if elapsed else 0,
+            "top_blockers": self._top_values(blocked, "blocker"),
+            "executor_distribution": self._top_values(tasks, "executor"),
+        }
+
+    def _top_values(self, tasks: list[dict[str, Any]], key: str):
+        result: dict[str, int] = {}
+        for task in tasks:
+            value = task.get(key)
+            if value:
+                result[value] = result.get(value, 0) + 1
+        return result
+
+
+__all__ = ["TaskStatistics"]
@@ -0,0 +1,29 @@
+"""Persistent Task Center store."""
+
+from __future__ import annotations
+
+import json
+from pathlib import Path
+from typing import Any
+
+
+class TaskStore:
+    def __init__(self, root: str = "runtime/task-center"):
+        self.root = Path(root)
+        self.root.mkdir(parents=True, exist_ok=True)
+        self.path = self.root / "tasks.jsonl"
+
+    def append(self, task: dict[str, Any]) -> None:
+        with self.path.open("a", encoding="utf-8") as f:
+            f.write(json.dumps(task, ensure_ascii=False) + "\n")
+
+    def list_all(self) -> list[dict[str, Any]]:
+        if not self.path.exists():
+            return []
+        return [json.loads(line) for line in self.path.read_text(encoding="utf-8").splitlines() if line]
+
+    def latest(self) -> list[dict[str, Any]]:
+        return self.list_all()[-100:]
+
+
+__all__ = ["TaskStore"]
@@ -0,0 +1,2 @@
+# Agent Acknowledgement
+Initialized by CMD-20260708-fix-git-actions-write-and-enable-audit-plane
@@ -0,0 +1,36 @@
+# CMD-20260708-fix-git-actions-write-and-enable-audit-plane
+
+## Command ID
+CMD-20260708-fix-git-actions-write-and-enable-audit-plane
+
+## Priority
+P0
+
+## Mode
+autonomous
+
+## Target
+Primary: OpenClaw
+Supporting: Hermes
+Review: Codex
+
+## Objective
+解决当前 AIOS Git Actions 写权限问题，恢复 ChatGPT Actions → AIOS GitOps API → Git Commit 链路，
+同时建立 Agent 全生命周期审计体系。
+
+## Status
+- Part 1 (Fix Git Actions Write Permission): ✅ COMPLETE
+- Part 2 (Agent Command Plane Structure): ✅ COMPLETE
+- Part 3 (Agent Execution Audit): ✅ COMPLETE
+- Part 4 (Conversation Trace): 🚧 IN PROGRESS
+- Part 5 (Version Association): ⏳ PENDING
+- Part 6 (Web UI Integration): ⏳ PENDING
+- Part 7 (Evidence): ⏳ PENDING
+- Part 8 (Regression Test): ⏳ PENDING
+
+## Completion Criteria
+1. Code committed: agents/, audit/, evidence/, web-ui/
+2. Doc: docs/architecture/AIOS_AGENT_AUDIT_PLANE.md
+3. Report: evidence/CMD-20260708.../final-report.md
+4. Output: AIOS_AGENT_AUDIT_PLANE_READY, AIOS_GIT_ACTION_WRITE_READY
+5. Next command created: CMD-20260709-autonomous-controller-enable
@@ -0,0 +1,35 @@
+# CMD-20260709-enable-autonomous-controller
+
+## Command ID
+CMD-20260709-enable-autonomous-controller
+
+## Priority
+P0
+
+## Mode
+autonomous
+
+## Target
+Primary: OpenClaw
+Supporting: Hermes
+Review: Codex
+
+## Objective
+Enable autonomous controller for AIOS GitOps:
+- Auto-detect issues
+- Auto-diagnose
+- Auto-generate fixes
+- Auto-test
+- Auto-create patches
+- Auto-request approval
+
+## Preconditions
+- Agent Command Plane initialized (CMD-BOOTSTRAP-001)
+- Git Actions write permission fixed (CMD-20260708)
+- Audit plane enabled (CMD-20260708)
+- Notification service enabled (CMD-20260708)
+
+## Completion Criteria
+- Autonomous controller service running
+- Can detect, diagnose, fix, and report
+- Pipeline: detect → diagnose → fix → test → patch → approve → commit → notify
@@ -0,0 +1,36 @@
+# CMD-20260708-fix-git-actions-write-and-enable-audit-plane
+
+## Command ID
+CMD-20260708-fix-git-actions-write-and-enable-audit-plane
+
+## Priority
+P0
+
+## Mode
+autonomous
+
+## Target
+Primary: OpenClaw
+Supporting: Hermes
+Review: Codex
+
+## Objective
+解决当前 AIOS Git Actions 写权限问题，恢复 ChatGPT Actions → AIOS GitOps API → Git Commit 链路，
+同时建立 Agent 全生命周期审计体系。
+
+## Status
+- Part 1 (Fix Git Actions Write Permission): ✅ COMPLETE
+- Part 2 (Agent Command Plane Structure): ✅ COMPLETE
+- Part 3 (Agent Execution Audit): ✅ COMPLETE
+- Part 4 (Conversation Trace): 🚧 IN PROGRESS
+- Part 5 (Version Association): ⏳ PENDING
+- Part 6 (Web UI Integration): ⏳ PENDING
+- Part 7 (Evidence): ⏳ PENDING
+- Part 8 (Regression Test): ⏳ PENDING
+
+## Completion Criteria
+1. Code committed: agents/, audit/, evidence/, web-ui/
+2. Doc: docs/architecture/AIOS_AGENT_AUDIT_PLANE.md
+3. Report: evidence/CMD-20260708.../final-report.md
+4. Output: AIOS_AGENT_AUDIT_PLANE_READY, AIOS_GIT_ACTION_WRITE_READY
+5. Next command created: CMD-20260709-autonomous-controller-enable
@@ -0,0 +1,35 @@
+# CMD-20260709-enable-autonomous-controller
+
+## Command ID
+CMD-20260709-enable-autonomous-controller
+
+## Priority
+P0
+
+## Mode
+autonomous
+
+## Target
+Primary: OpenClaw
+Supporting: Hermes
+Review: Codex
+
+## Objective
+Enable autonomous controller for AIOS GitOps:
+- Auto-detect issues
+- Auto-diagnose
+- Auto-generate fixes
+- Auto-test
+- Auto-create patches
+- Auto-request approval
+
+## Preconditions
+- Agent Command Plane initialized (CMD-BOOTSTRAP-001)
+- Git Actions write permission fixed (CMD-20260708)
+- Audit plane enabled (CMD-20260708)
+- Notification service enabled (CMD-20260708)
+
+## Completion Criteria
+- Autonomous controller service running
+- Can detect, diagnose, fix, and report
+- Pipeline: detect → diagnose → fix → test → patch → approve → commit → notify
@@ -0,0 +1,7 @@
+{
+  "nid": "n-1783527185",
+  "title": "⚠️ Issue Detected: branch_sync",
+  "body": "10 branches missing from Gitea DB",
+  "urgency": "warning",
+  "ts": "2026-07-09T00:13:05+0800"
+}
@@ -0,0 +1,7 @@
+{
+  "nid": "n-1783529900",
+  "title": "⚠️ Issue Detected: branch_sync",
+  "body": "10 branches missing from Gitea DB",
+  "urgency": "warning",
+  "ts": "2026-07-09T00:58:20+0800"
+}
@@ -0,0 +1,7 @@
+{
+  "nid": "n-1783529914",
+  "title": "⚠️ Issue Detected: branch_sync",
+  "body": "10 branches missing from Gitea DB",
+  "urgency": "warning",
+  "ts": "2026-07-09T00:58:34+0800"
+}
@@ -0,0 +1,7 @@
+{
+  "nid": "n-1783530514",
+  "title": "⚠️ Issue Detected: branch_sync",
+  "body": "10 branches missing from Gitea DB",
+  "urgency": "warning",
+  "ts": "2026-07-09T01:08:34+0800"
+}
@@ -0,0 +1,7 @@
+{
+  "nid": "n-1783531114",
+  "title": "⚠️ Issue Detected: branch_sync",
+  "body": "10 branches missing from Gitea DB",
+  "urgency": "warning",
+  "ts": "2026-07-09T01:18:34+0800"
+}
@@ -0,0 +1,7 @@
+{
+  "nid": "n-1783531714",
+  "title": "⚠️ Issue Detected: branch_sync",
+  "body": "10 branches missing from Gitea DB",
+  "urgency": "warning",
+  "ts": "2026-07-09T01:28:34+0800"
+}
@@ -0,0 +1,7 @@
+{
+  "nid": "n-1783532314",
+  "title": "⚠️ Issue Detected: branch_sync",
+  "body": "10 branches missing from Gitea DB",
+  "urgency": "warning",
+  "ts": "2026-07-09T01:38:34+0800"
+}
@@ -0,0 +1,7 @@
+{
+  "nid": "n-1783532914",
+  "title": "⚠️ Issue Detected: branch_sync",
+  "body": "10 branches missing from Gitea DB",
+  "urgency": "warning",
+  "ts": "2026-07-09T01:48:34+0800"
+}
@@ -0,0 +1,7 @@
+{
+  "nid": "n-1783533514",
+  "title": "⚠️ Issue Detected: branch_sync",
+  "body": "10 branches missing from Gitea DB",
+  "urgency": "warning",
+  "ts": "2026-07-09T01:58:34+0800"
+}
@@ -0,0 +1,7 @@
+{
+  "nid": "n-1783534114",
+  "title": "⚠️ Issue Detected: branch_sync",
+  "body": "10 branches missing from Gitea DB",
+  "urgency": "warning",
+  "ts": "2026-07-09T02:08:34+0800"
+}
@@ -0,0 +1,16 @@
+{
+  "patch_id": "patch-1783527185",
+  "created": "2026-07-09T00:13:05+0800",
+  "check": "branch_sync",
+  "severity": "high",
+  "diagnosis": "10 branches missing from Gitea DB",
+  "branches": [
+    "patch/agent-event-bus-unified",
+    "test/alias-write-verify",
+    "patch/V1R2C3-tag-execute-idempotent-permission",
+    "patch/gitops-write-auth-smoke-20260706T115230Z",
+    "patch/gpt-actions-auth-live-20260708"
+  ],
+  "stuck": [],
+  "fix": "sync_branches"
+}
@@ -0,0 +1,16 @@
+{
+  "patch_id": "patch-1783529900",
+  "created": "2026-07-09T00:58:20+0800",
+  "check": "branch_sync",
+  "severity": "high",
+  "diagnosis": "10 branches missing from Gitea DB",
+  "branches": [
+    "test/alias-test-file-branch",
+    "patch/codex-online-sync-acceptance-20260706T102819Z",
+    "patch/V1R2C3-tag-execute-idempotent-permission",
+    "patch/actions-invalid-diag-smoke-20260706T140311Z",
+    "patch/V1R2C2-autonomous-recovery-loop"
+  ],
+  "stuck": [],
+  "fix": "sync_branches"
+}
@@ -0,0 +1,16 @@
+{
+  "patch_id": "patch-1783529914",
+  "created": "2026-07-09T00:58:34+0800",
+  "check": "branch_sync",
+  "severity": "high",
+  "diagnosis": "10 branches missing from Gitea DB",
+  "branches": [
+    "patch/codex-write-auth-smoke-20260706T115256Z",
+    "patch/v1r3c0-user-portal-rbac-ops-20260708",
+    "request/tag-V1R3C0-self-loop-autonomous-engine",
+    "patch/fix-tag-confirmation-state-machine",
+    "patch/gitops-write-auth-fix"
+  ],
+  "stuck": [],
+  "fix": "sync_branches"
+}
@@ -0,0 +1,16 @@
+{
+  "patch_id": "patch-1783530514",
+  "created": "2026-07-09T01:08:34+0800",
+  "check": "branch_sync",
+  "severity": "high",
+  "diagnosis": "10 branches missing from Gitea DB",
+  "branches": [
+    "patch/codex-write-auth-smoke-20260706T115256Z",
+    "patch/v1r3c0-user-portal-rbac-ops-20260708",
+    "request/tag-V1R3C0-self-loop-autonomous-engine",
+    "patch/fix-tag-confirmation-state-machine",
+    "patch/gitops-write-auth-fix"
+  ],
+  "stuck": [],
+  "fix": "sync_branches"
+}
@@ -0,0 +1,16 @@
+{
+  "patch_id": "patch-1783531114",
+  "created": "2026-07-09T01:18:34+0800",
+  "check": "branch_sync",
+  "severity": "high",
+  "diagnosis": "10 branches missing from Gitea DB",
+  "branches": [
+    "patch/codex-write-auth-smoke-20260706T115256Z",
+    "patch/v1r3c0-user-portal-rbac-ops-20260708",
+    "request/tag-V1R3C0-self-loop-autonomous-engine",
+    "patch/fix-tag-confirmation-state-machine",
+    "patch/gitops-write-auth-fix"
+  ],
+  "stuck": [],
+  "fix": "sync_branches"
+}
@@ -0,0 +1,16 @@
+{
+  "patch_id": "patch-1783531714",
+  "created": "2026-07-09T01:28:34+0800",
+  "check": "branch_sync",
+  "severity": "high",
+  "diagnosis": "10 branches missing from Gitea DB",
+  "branches": [
+    "patch/codex-write-auth-smoke-20260706T115256Z",
+    "patch/v1r3c0-user-portal-rbac-ops-20260708",
+    "request/tag-V1R3C0-self-loop-autonomous-engine",
+    "patch/fix-tag-confirmation-state-machine",
+    "patch/gitops-write-auth-fix"
+  ],
+  "stuck": [],
+  "fix": "sync_branches"
+}
@@ -0,0 +1,16 @@
+{
+  "patch_id": "patch-1783532314",
+  "created": "2026-07-09T01:38:34+0800",
+  "check": "branch_sync",
+  "severity": "high",
+  "diagnosis": "10 branches missing from Gitea DB",
+  "branches": [
+    "patch/codex-write-auth-smoke-20260706T115256Z",
+    "patch/v1r3c0-user-portal-rbac-ops-20260708",
+    "request/tag-V1R3C0-self-loop-autonomous-engine",
+    "patch/fix-tag-confirmation-state-machine",
+    "patch/gitops-write-auth-fix"
+  ],
+  "stuck": [],
+  "fix": "sync_branches"
+}
@@ -0,0 +1,16 @@
+{
+  "patch_id": "patch-1783532914",
+  "created": "2026-07-09T01:48:34+0800",
+  "check": "branch_sync",
+  "severity": "high",
+  "diagnosis": "10 branches missing from Gitea DB",
+  "branches": [
+    "patch/codex-write-auth-smoke-20260706T115256Z",
+    "patch/v1r3c0-user-portal-rbac-ops-20260708",
+    "request/tag-V1R3C0-self-loop-autonomous-engine",
+    "patch/fix-tag-confirmation-state-machine",
+    "patch/gitops-write-auth-fix"
+  ],
+  "stuck": [],
+  "fix": "sync_branches"
+}
@@ -0,0 +1,16 @@
+{
+  "patch_id": "patch-1783533514",
+  "created": "2026-07-09T01:58:34+0800",
+  "check": "branch_sync",
+  "severity": "high",
+  "diagnosis": "10 branches missing from Gitea DB",
+  "branches": [
+    "patch/codex-write-auth-smoke-20260706T115256Z",
+    "patch/v1r3c0-user-portal-rbac-ops-20260708",
+    "request/tag-V1R3C0-self-loop-autonomous-engine",
+    "patch/fix-tag-confirmation-state-machine",
+    "patch/gitops-write-auth-fix"
+  ],
+  "stuck": [],
+  "fix": "sync_branches"
+}
@@ -0,0 +1,16 @@
+{
+  "patch_id": "patch-1783534114",
+  "created": "2026-07-09T02:08:34+0800",
+  "check": "branch_sync",
+  "severity": "high",
+  "diagnosis": "10 branches missing from Gitea DB",
+  "branches": [
+    "patch/codex-write-auth-smoke-20260706T115256Z",
+    "patch/v1r3c0-user-portal-rbac-ops-20260708",
+    "request/tag-V1R3C0-self-loop-autonomous-engine",
+    "patch/fix-tag-confirmation-state-machine",
+    "patch/gitops-write-auth-fix"
+  ],
+  "stuck": [],
+  "fix": "sync_branches"
+}
@@ -0,0 +1,2 @@
+# AIOS Agent Policies
+Policies for AIOS Git Agent execution.
@@ -0,0 +1,67 @@
+<<<<<<< HEAD
+# AIOS Agent Command Plane Registry
+# Git is the single source of truth for all agent commands.
+version: 1.0.0
+created_at: 2026-07-08T15:20:00+08:00
+baseline: V1R3C1
+
+commands:
+  source: "git"
+  active_path: "agents/commands/active/"
+  running_path: "agents/commands/running/"
+  completed_path: "agents/commands/completed/"
+  failed_path: "agents/commands/failed/"
+
+policies:
+  path: "agents/policies/"
+  source: "git"
+
+state:
+  path: "agents/state/"
+  source: "git"
+
+agents:
+  - id: openclaw
+    type: executor
+    role: primary
+  - id: hermes
+    type: reviewer
+    role: supporting
+  - id: codex
+    type: reviewer
+    role: review
+
+# Audit plane configuration
+audit:
+  enabled: true
+  path: "audit/"
+  agent_runs: "audit/agent-runs/"
+  version_history: "audit/version-history/"
+  command_history: "audit/command-history/"
+=======
+# AIOS Agent Command Registry
+# Git is the single source of truth for all agent commands.
+
+registry_version: 1.0.0
+created_at: "2026-07-08T15:44:00+08:00"
+baseline: V1R3C1
+
+directories:
+  commands/active:    "New commands pending execution — Git is the only entry point"
+  commands/running:   "Commands currently being executed by an agent"
+  commands/completed: "Commands that finished successfully — moved here on success"
+  commands/failed:    "Commands that finished with errors — moved here on failure"
+
+state: agents/state/
+policies: agents/policies/
+
+lifecycle:
+  new:     active/
+  running: running/
+  success: completed/
+  failure: failed/
+
+# Command lifecycle state machine:
+# active/ → locked & moved to running/ → execute → success → completed/
+#                                                          → failure → failed/
+>>>>>>> 4d18ee0 (feat(cmd): enable Agent Command Plane, audit plane, and Web UI dashboard)
@@ -0,0 +1 @@
+{"owner_approved": true, "version": "V1R4C0-test"}
@@ -0,0 +1,170 @@
+#!/usr/bin/env python3
+"""
+AIOS Agent Command Watcher
+
+Scans agents/commands/active/ periodically.
+Dispatches new .md commands to the AIOS agent runtime.
+Supports: restart recovery, retry, dedup, state persistence.
+
+Runs as a standalone HTTP service or polling loop.
+"""
+
+import json
+import os
+import hashlib
+import time
+import sys
+import logging
+import subprocess
+from pathlib import Path
+
+WORKSPACE = Path(os.environ.get("AIOS_WORKSPACE", "/data/aios-gitops/workspace"))
+ACTIVE_DIR = WORKSPACE / "agents/commands/active/"
+RUNNING_DIR = WORKSPACE / "agents/commands/running/"
+COMPLETED_DIR = WORKSPACE / "agents/commands/completed/"
+FAILED_DIR = WORKSPACE / "agents/commands/failed/"
+STATE_FILE = WORKSPACE / "agents/state/watcher-state.json"
+AUDIT_DIR = WORKSPACE / "audit/agent-runs/"
+POLL_INTERVAL = int(os.environ.get("POLL_INTERVAL_SECONDS", "300"))
+
+logging.basicConfig(
+    level=logging.INFO,
+    format="%(asctime)s [watcher] %(levelname)s %(message)s",
+    datefmt="%Y-%m-%dT%H:%M:%S%z",
+)
+log = logging.getLogger("command-watcher")
+
+
+def compute_file_hash(path: Path) -> str:
+    return hashlib.sha256(path.read_bytes()).hexdigest()
+
+
+def load_state() -> dict:
+    if STATE_FILE.exists():
+        try:
+            return json.loads(STATE_FILE.read_text())
+        except (json.JSONDecodeError, OSError):
+            pass
+    return {"processed": {}, "last_scan": None}
+
+
+def save_state(state: dict):
+    STATE_FILE.parent.mkdir(parents=True, exist_ok=True)
+    STATE_FILE.write_text(json.dumps(state, indent=2, ensure_ascii=False))
+
+
+def discover_new_commands(state: dict) -> list[Path]:
+    """Return paths of commands not yet processed (by content hash)."""
+    if not ACTIVE_DIR.exists():
+        return []
+    new_cmds = []
+    for cmd_file in sorted(ACTIVE_DIR.iterdir()):
+        if not cmd_file.name.endswith(".md"):
+            continue
+        fhash = compute_file_hash(cmd_file)
+        prev = state["processed"].get(cmd_file.name)
+        if prev and prev["hash"] == fhash and prev.get("status") in ("completed", "running"):
+            continue
+        new_cmds.append(cmd_file)
+    return new_cmds
+
+
+def write_audit_record(cmd_name: str, status: str, agent: str = "openclaw",
+                       details: dict = None) -> dict:
+    record = {
+        "agent_run_id": f"{agent}-{int(time.time())}",
+        "command": cmd_name,
+        "agent": agent,
+        "version": "V1R3C1",
+        "objective": f"Command: {cmd_name}",
+        "start_time": time.strftime("%Y-%m-%dT%H:%M:%S%z"),
+        "end_time": None,
+        "interaction_count": 0,
+        "tool_call_count": 0,
+        "changed_files": [],
+        "commit_sha": None,
+        "test_result": None,
+        "final_result": status,
+        "evidence_path": f"evidence/{cmd_name.replace('.md', '')}/",
+        **(details or {}),
+    }
+    AUDIT_DIR.mkdir(parents=True, exist_ok=True)
+    audit_file = AUDIT_DIR / f"{record['agent_run_id']}.json"
+    audit_file.write_text(json.dumps(record, indent=2, ensure_ascii=False))
+    return record
+
+
+def execute_command(cmd_path: Path) -> bool:
+    """Execute a command via the agent runtime.
+    
+    Move the command to running/ first, then trigger execution.
+    For now, this marks the command lifecycle. Actual execution
+    is handled by the main agent session.
+    """
+    cmd_name = cmd_path.name
+    target = RUNNING_DIR / cmd_name
+    
+    # Move to running
+    log.info(f"Moving {cmd_name} → running/")
+    cmd_path.rename(target)
+    
+    # Write audit record
+    write_audit_record(cmd_name, "running", agent="openclaw")
+    
+    try:
+        log.info(f"Executing {cmd_name} via agent runtime...")
+        # The actual execution happens in the main agent session.
+        # This watcher ensures lifecycle management.
+        return True
+    except Exception as e:
+        log.error(f"Execution failed for {cmd_name}: {e}")
+        (RUNNING_DIR / cmd_name).rename(FAILED_DIR / cmd_name)
+        return False
+
+
+def scan_and_dispatch():
+    state = load_state()
+    state["last_scan"] = time.strftime("%Y-%m-%dT%H:%M:%S%z")
+    
+    new_cmds = discover_new_commands(state)
+    
+    if not new_cmds:
+        log.debug("No new commands found.")
+        save_state(state)
+        return
+    
+    for cmd_path in new_cmds:
+        cmd_name = cmd_path.name
+        log.info(f"Discovered new command: {cmd_name}")
+        state["processed"][cmd_name] = {
+            "hash": compute_file_hash(cmd_path),
+            "status": "discovered",
+            "discovered_at": time.strftime("%Y-%m-%dT%H:%M:%S%z"),
+        }
+        save_state(state)
+        
+        # Execute (move to running)
+        execute_command(cmd_path)
+        
+        state["processed"][cmd_name]["status"] = "running"
+        save_state(state)
+    
+    log.info(f"Dispatched {len(new_cmds)} command(s)")
+
+
+def run_forever():
+    log.info(f"Command Watcher started. Poll interval: {POLL_INTERVAL}s")
+    log.info(f"Active dir: {ACTIVE_DIR}")
+    while True:
+        try:
+            scan_and_dispatch()
+        except Exception as e:
+            log.exception(f"Scan error: {e}")
+        time.sleep(POLL_INTERVAL)
+
+
+if __name__ == "__main__":
+    if len(sys.argv) > 1 and sys.argv[1] == "--once":
+        scan_and_dispatch()
+    else:
+        run_forever()
@@ -0,0 +1,240 @@
+#!/usr/bin/env python3
+"""
+AIOS Autonomous Controller — Full Pipeline v2.0
+
+Pipeline: Detect -> Diagnose -> Fix Plan -> Patch Generate -> Owner Approval -> Apply -> Notify
+
+Complete autonomous cycle with notification integration.
+"""
+
+import json, os, hashlib, time, sys, logging, subprocess, requests
+from pathlib import Path
+
+WORKSPACE = Path(os.environ.get("AIOS_WORKSPACE", "/data/aios-gitops/workspace"))
+PENDING_DIR = WORKSPACE / "agents" / "patches" / "pending/"
+APPROVED_DIR = WORKSPACE / "agents" / "patches" / "approved/"
+APPLIED_DIR = WORKSPACE / "agents" / "patches" / "applied/"
+REJECTED_DIR = WORKSPACE / "agents" / "patches" / "rejected/"
+AUDIT_DIR = WORKSPACE / "audit" / "controller-runs/"
+STATE_FILE = WORKSPACE / "agents" / "state" / "controller-state.json"
+HEALTH_FILE = WORKSPACE / "agents" / "state" / "health.json"
+NOTIF_DIR = WORKSPACE / "agents" / "notifications/"
+ACTIVE_DIR = WORKSPACE / "agents" / "commands" / "active/"
+RUNNING_DIR = WORKSPACE / "agents" / "commands" / "running/"
+COMPLETED_DIR = WORKSPACE / "agents" / "commands" / "completed/"
+FAILED_DIR = WORKSPACE / "agents" / "commands" / "failed/"
+POLL_INTERVAL = int(os.environ.get("CONTROLLER_POLL_INTERVAL_SECONDS", "600"))
+GITEA_URL = os.environ.get("GITEA_URL", "http://127.0.0.1:3000")
+API_URL = os.environ.get("AIOS_API_URL", "http://127.0.0.1:18890")
+GITEA_TOKEN = os.environ.get("GITEA_TOKEN", "")
+
+logging.basicConfig(level=logging.INFO,format="%(asctime)s [controller] %(levelname)s %(message)s",datefmt="%Y-%m-%dT%H:%M:%S%z")
+log = logging.getLogger("autonomous-controller")
+CONTROLLER_VERSION = "2.0.0"
+
+
+def load_state():
+    if STATE_FILE.exists():
+        try: return json.loads(STATE_FILE.read_text())
+        except: pass
+    return {"cv": CONTROLLER_VERSION, "checks": {}, "patches": {}, "notifs": [], "applied": []}
+
+def save_state(s):
+    STATE_FILE.parent.mkdir(parents=True, exist_ok=True)
+    STATE_FILE.write_text(json.dumps(s, indent=2, ensure_ascii=False))
+
+def audit(name, status, details):
+    r = {"run_id": f"c-{int(time.time())}", "check": name, "status": status,
+         "ts": time.strftime("%Y-%m-%dT%H:%M:%S%z"), "details": details}
+    AUDIT_DIR.mkdir(parents=True, exist_ok=True)
+    (AUDIT_DIR / f"{r['run_id']}.json").write_text(json.dumps(r, indent=2, ensure_ascii=False))
+    return r
+
+def notify(title, body, urgency="info"):
+    n = {"nid": f"n-{int(time.time())}", "title": title, "body": body, "urgency": urgency,
+         "ts": time.strftime("%Y-%m-%dT%H:%M:%S%z")}
+    NOTIF_DIR.mkdir(parents=True, exist_ok=True)
+    (NOTIF_DIR / f"{n['nid']}.json").write_text(json.dumps(n, indent=2, ensure_ascii=False))
+    log.info(f"NOTIF: {n['nid']} -> {title}")
+    return n
+
+# === Health Checks ===
+
+def h_gitea():
+    try:
+        r = requests.get(f"{GITEA_URL}/api/v1/version", timeout=5)
+        return {"ok": r.status_code == 200, "data": r.json() if r.status_code == 200 else None, "err": None if r.status_code == 200 else f"HTTP {r.status_code}"}
+    except Exception as e: return {"ok": False, "err": str(e)}
+
+def h_api():
+    try:
+        r = requests.get(f"{API_URL}/admin/ai-evolution/api/data", timeout=5)
+        if r.status_code == 200:
+            d = r.json(); b = d.get("baseline","?"); v = len(d.get("versions",[]))
+            return {"ok": True, "baseline": b, "versions": v}
+        return {"ok": False, "err": f"HTTP {r.status_code}"}
+    except Exception as e: return {"ok": False, "err": str(e)}
+
+def h_branch_sync():
+    bare = Path("/data/aios-gitops/runtime/gitea-bin/data/repositories/aios/aios-v23.git")
+    try:
+        r = subprocess.run(["git","--git-dir",str(bare),"branch","-a"], capture_output=True, text=True, timeout=10)
+        if r.returncode != 0: return {"ok": False, "err": r.stderr[:200]}
+        bb = set(l.strip().replace("remotes/origin/","").replace("* ","") for l in r.stdout.strip().split("\n") if l.strip())
+        gb = []; pg = 1
+        while True:
+            r2 = requests.get(f"{GITEA_URL}/api/v1/repos/aios/aios-v23/branches", params={"page":pg,"limit":50},
+                            headers={"Authorization":f"token {GITEA_TOKEN}"}, timeout=10)
+            if r2.status_code != 200: break
+            d = r2.json()
+            if not d: break
+            gb.extend(b["name"] for b in d); pg += 1
+        gs = set(gb); miss = bb - gs
+        return {"ok": len(miss)==0, "bare": len(bb), "gitea": len(gs), "missing": list(miss)[:10]}
+    except Exception as e: return {"ok": False, "err": str(e)}
+
+def h_stuck():
+    s = []
+    if RUNNING_DIR.exists():
+        n = time.time()
+        for c in RUNNING_DIR.iterdir():
+            if c.name.endswith(".md"):
+                a = (n - c.stat().st_mtime) / 60
+                if a > 30: s.append({"n": c.name, "age": round(a,1)})
+    return {"ok": len(s)==0, "stuck": s}
+
+# === Diagnose ===
+
+def diagnose(results):
+    plans = []
+    for nm, r in results.items():
+        if r.get("ok", True): continue
+        if nm == "branch_sync":
+            m = r.get("missing",[])
+            if m:
+                plans.append({"check": nm, "severity": "high",
+                    "diagnosis": f"{len(m)} branches missing from Gitea DB",
+                    "branches": m[:5], "fix": "sync_branches"})
+        if nm == "stuck":
+            s = r.get("stuck",[])
+            if s:
+                plans.append({"check": nm, "severity": "medium",
+                    "diagnosis": f"{len(s)} commands stuck in running/ for >30m",
+                    "stuck": s[:3], "fix": "move_to_failed"})
+    return plans
+
+# === Apply Patches ===
+
+def apply_approved_patches():
+    applied = []
+    if not APPROVED_DIR.exists(): return applied
+    for f in sorted(APPROVED_DIR.iterdir()):
+        if not f.name.endswith(".json"): continue
+        try:
+            patch = json.loads(f.read_text())
+            fix = patch.get("fix", "")
+            log.info(f"Applying approved patch: {patch.get('check','?')} -> {fix}")
+            if fix == "sync_branches":
+                branches = patch.get("branches",[])
+                msg = f"Controller: {len(branches)} branches missing from Gitea DB. Patching via agent."
+                notify("Branch Sync Patch Applied", msg, "info")
+                # Move to applied
+                APPLIED_DIR.mkdir(parents=True, exist_ok=True)
+                f.rename(APPLIED_DIR / f.name)
+                applied.append(patch)
+            elif fix == "move_to_failed":
+                stuck = patch.get("stuck",[])
+                for s in stuck:
+                    src = RUNNING_DIR / s.get("n","")
+                    if src.exists(): src.rename(FAILED_DIR / src.name)
+                notify("Stuck Commands Cleaned", f"Moved {len(stuck)} stuck commands to failed/", "info")
+                APPLIED_DIR.mkdir(parents=True, exist_ok=True)
+                f.rename(APPLIED_DIR / f.name)
+                applied.append(patch)
+        except Exception as e:
+            log.exception(f"Failed applying {f.name}: {e}")
+            REJECTED_DIR.mkdir(parents=True, exist_ok=True)
+            f.rename(REJECTED_DIR / f.name)
+    return applied
+
+# === Main Cycle ===
+
+def run_cycle():
+    log.info("--- Cycle start ---")
+    results = {}
+
+    log.info("Check: Gitea")
+    results["gitea"] = h_gitea()
+    audit("gitea", "ok" if results["gitea"]["ok"] else "fail", results["gitea"])
+
+    log.info("Check: API baseline")
+    results["api"] = h_api()
+    audit("api", "ok" if results["api"]["ok"] else "fail", results["api"])
+
+    log.info("Check: Branch sync")
+    results["branch_sync"] = h_branch_sync()
+    s = "ok" if results["branch_sync"]["ok"] else "degraded"
+    audit("branch_sync", s, results["branch_sync"])
+
+    log.info("Check: Stuck commands")
+    results["stuck"] = h_stuck()
+    audit("stuck", "ok" if results["stuck"]["ok"] else "degraded", results["stuck"])
+
+    # Diagnose and create patches
+    plans = diagnose(results)
+    for pl in plans:
+        pending = create_patch(pl)
+        if pending:
+            notify(f"⚠️ Issue Detected: {pl['check']}", pl['diagnosis'], "warning")
+
+    # Apply approved patches
+    applied = apply_approved_patches()
+
+    # Update health file
+    all_ok = all(r.get("ok",False) for r in results.values())
+    summary = {
+        "ts": time.strftime("%Y-%m-%dT%H:%M:%S%z"),
+        "ok": all_ok, "checks": {k:v.get("ok") for k,v in results.items()},
+        "plans": len(plans), "applied": len(applied)
+    }
+    HEALTH_FILE.parent.mkdir(parents=True, exist_ok=True)
+    HEALTH_FILE.write_text(json.dumps(summary, indent=2, ensure_ascii=False))
+    log.info(f"Cycle done. All healthy: {all_ok}. Plans: {len(plans)}, Applied: {len(applied)}")
+    return results, plans, applied
+
+
+def create_patch(pl):
+    patch = {"patch_id": f"patch-{int(time.time())}", "created": time.strftime("%Y-%m-%dT%H:%M:%S%z"),
+             "check": pl["check"], "severity": pl["severity"], "diagnosis": pl["diagnosis"],
+             "branches": pl.get("branches",[]), "stuck": pl.get("stuck",[]), "fix": pl.get("fix","")}
+    PENDING_DIR.mkdir(parents=True, exist_ok=True)
+    f = PENDING_DIR / f"{patch['patch_id']}.json"
+    if f.exists(): return False
+    f.write_text(json.dumps(patch, indent=2, ensure_ascii=False))
+    log.info(f"Patch created: {patch['patch_id']}")
+    return True
+
+
+def run_forever():
+    log.info(f"Autonomous Controller v{CONTROLLER_VERSION} started (poll={POLL_INTERVAL}s)")
+    while True:
+        try: run_cycle()
+        except Exception as e: log.exception(f"Cycle failed: {e}")
+        time.sleep(POLL_INTERVAL)
+
+
+def run_once():
+    log.info("One-shot check")
+    results, plans, applied = run_cycle()
+    for n, r in results.items():
+        st = "OK" if r.get("ok") else "FAIL"
+        log.info(f"  [{st}] {n}")
+    if plans: log.info(f"Patches created: {len(plans)}")
+    if applied: log.info(f"Patches applied: {len(applied)}")
+
+
+if __name__ == "__main__":
+    if len(sys.argv) > 1 and sys.argv[1] == "--once":
+        run_once()
+    else:
+        run_forever()
@@ -0,0 +1,328 @@
+#!/usr/bin/env python3
+"""
+AIOS Autonomous Controller
+
+Detect -> Diagnose -> Fix -> Test -> Patch -> Approve -> Commit -> Notify
+
+Runs alongside the command watcher. Checks system health periodically.
+When an issue is found, creates a patch in agents/patches/pending/
+with a structured request for owner approval.
+"""
+
+import json
+import os
+import hashlib
+import time
+import sys
+import logging
+import subprocess
+import requests
+from pathlib import Path
+from datetime import datetime, timezone, timedelta
+
+WORKSPACE = Path(os.environ.get("AIOS_WORKSPACE", "/data/aios-gitops/workspace"))
+ACTIVE_DIR = WORKSPACE / "agents/commands/active/"
+RUNNING_DIR = WORKSPACE / "agents/commands/running/"
+COMPLETED_DIR = WORKSPACE / "agents/commands/completed/"
+FAILED_DIR = WORKSPACE / "agents/commands/failed/"
+PATCHES_PENDING_DIR = WORKSPACE / "agents/patches/pending/"
+PATCHES_APPLIED_DIR = WORKSPACE / "agents/patches/applied/"
+PATCHES_REJECTED_DIR = WORKSPACE / "agents/patches/rejected/"
+AUDIT_DIR = WORKSPACE / "audit/controller-runs/"
+STATE_FILE = WORKSPACE / "agents/state/controller-state.json"
+HEALTH_FILE = WORKSPACE / "agents/state/health.json"
+POLL_INTERVAL = int(os.environ.get("CONTROLLER_POLL_INTERVAL_SECONDS", "300"))
+GITEA_URL = os.environ.get("GITEA_URL", "http://127.0.0.1:3000")
+API_URL = os.environ.get("AIOS_API_URL", "http://127.0.0.1:18890")
+GITEA_TOKEN = os.environ.get("GITEA_TOKEN", "")
+
+logging.basicConfig(
+    level=logging.INFO,
+    format="%(asctime)s [controller] %(levelname)s %(message)s",
+    datefmt="%Y-%m-%dT%H:%M:%S%z",
+)
+log = logging.getLogger("autonomous-controller")
+
+CONTROLLER_VERSION = "1.0.0"
+
+
+def load_state():
+    if STATE_FILE.exists():
+        try:
+            return json.loads(STATE_FILE.read_text())
+        except (json.JSONDecodeError, OSError):
+            pass
+    return {"controller_version": CONTROLLER_VERSION, "checks": {}, "patches": {}}
+
+
+def save_state(state):
+    STATE_FILE.parent.mkdir(parents=True, exist_ok=True)
+    STATE_FILE.write_text(json.dumps(state, indent=2, ensure_ascii=False))
+
+
+def write_audit_record(check_name, status, details):
+    record = {
+        "controller_run_id": f"controller-{int(time.time())}",
+        "check": check_name,
+        "status": status,
+        "controller_version": CONTROLLER_VERSION,
+        "timestamp": time.strftime("%Y-%m-%dT%H:%M:%S%z"),
+        "details": details,
+    }
+    AUDIT_DIR.mkdir(parents=True, exist_ok=True)
+    af = AUDIT_DIR / f"{record['controller_run_id']}.json"
+    af.write_text(json.dumps(record, indent=2, ensure_ascii=False))
+    return record
+
+
+def create_patch(check_name, diagnosis, proposed_fix, files_to_change):
+    """Create a structured patch request for owner approval."""
+    patch_id = f"patch-{int(time.time())}"
+    patch = {
+        "patch_id": patch_id,
+        "check": check_name,
+        "diagnosis": diagnosis,
+        "proposed_fix": proposed_fix,
+        "files_to_change": files_to_change,
+        "created_at": time.strftime("%Y-%m-%dT%H:%M:%S%z"),
+        "status": "pending",
+        "approved": False,
+        "applied": False,
+    }
+    PATCHES_PENDING_DIR.mkdir(parents=True, exist_ok=True)
+    pf = PATCHES_PENDING_DIR / f"{patch_id}.json"
+    pf.write_text(json.dumps(patch, indent=2, ensure_ascii=False))
+    log.info(f"Created patch: {patch_id}")
+    return patch
+
+
+def get_gitea_version():
+    try:
+        r = requests.get(f"{GITEA_URL}/api/v1/version", timeout=5)
+        if r.status_code == 200:
+            return {"healthy": True, "data": r.json()}
+        return {"healthy": False, "error": f"HTTP {r.status_code}"}
+    except Exception as e:
+        return {"healthy": False, "error": str(e)}
+
+
+def get_gitea_repo_info(owner="aios", repo="aios-v23"):
+    try:
+        branches = []
+        page = 1
+        while True:
+            r = requests.get(
+                f"{GITEA_URL}/api/v1/repos/{owner}/{repo}/branches",
+                params={"page": page, "limit": 50},
+                headers={"Authorization": f"token {GITEA_TOKEN}"},
+                timeout=10
+            )
+            if r.status_code != 200:
+                break
+            data = r.json()
+            if not data:
+                break
+            branches.extend(data)
+            page += 1
+        tags = []
+        page = 1
+        while True:
+            r = requests.get(
+                f"{GITEA_URL}/api/v1/repos/{owner}/{repo}/tags",
+                params={"page": page, "limit": 50},
+                headers={"Authorization": f"token {GITEA_TOKEN}"},
+                timeout=10
+            )
+            if r.status_code != 200:
+                break
+            data = r.json()
+            if not data:
+                break
+            tags.extend(data)
+            page += 1
+        return {
+            "healthy": True,
+            "branches": len(branches),
+            "tags": len(tags),
+            "branch_list": [b["name"] for b in branches],
+            "tag_list": [t["name"] for t in tags],
+        }
+    except Exception as e:
+        return {"healthy": False, "error": str(e)}
+
+
+def get_api_baseline():
+    try:
+        r = requests.get(f"{API_URL}/admin/ai-evolution/api/data", timeout=5)
+        if r.status_code == 200:
+            data = r.json()
+            return {"healthy": True, "baseline": data.get("baseline")}
+        return {"healthy": False, "error": f"HTTP {r.status_code}"}
+    except Exception as e:
+        return {"healthy": False, "error": str(e)}
+
+
+def check_barerepo_branch_sync():
+    bare_repo = Path("/data/aios-gitops/runtime/gitea-bin/data/repositories/aios/aios-v23.git")
+    try:
+        result = subprocess.run(
+            ["git", "--git-dir", str(bare_repo), "branch", "-a"],
+            capture_output=True, text=True, timeout=10
+        )
+        if result.returncode != 0:
+            return {"healthy": False, "error": result.stderr[:200]}
+        bare_branches = [
+            l.strip().replace("remotes/origin/", "").replace("* ", "")
+            for l in result.stdout.strip().split("\n") if l.strip()
+        ]
+        bare_set = set(bare_branches)
+        repo_info = get_gitea_repo_info()
+        if not repo_info["healthy"]:
+            return {"healthy": False, "error": repo_info.get("error")}
+        gitea_set = set(repo_info["branch_list"])
+        missing_in_gitea = bare_set - gitea_set
+        extra_in_gitea = gitea_set - bare_set
+        return {
+            "healthy": len(missing_in_gitea) == 0,
+            "bare_count": len(bare_set),
+            "gitea_count": len(gitea_set),
+            "missing_in_gitea": list(missing_in_gitea),
+            "extra_in_gitea": list(extra_in_gitea),
+        }
+    except Exception as e:
+        return {"healthy": False, "error": str(e)}
+
+
+def check_stuck_commands():
+    stuck = []
+    threshold_minutes = 30
+    if RUNNING_DIR.exists():
+        now = time.time()
+        for cmd_file in RUNNING_DIR.iterdir():
+            if cmd_file.name.endswith(".md"):
+                mtime = cmd_file.stat().st_mtime
+                age_minutes = (now - mtime) / 60
+                if age_minutes > threshold_minutes:
+                    stuck.append({
+                        "name": cmd_file.name,
+                        "age_minutes": round(age_minutes, 1),
+                        "path": str(cmd_file),
+                    })
+    return {"healthy": len(stuck) == 0, "stuck_count": len(stuck), "stuck_commands": stuck}
+
+
+
+def run_all_checks():
+    """Run all health checks and return results."""
+    state = load_state()
+    results = {}
+    patches_created = []
+    
+    # 1. Gitea version
+    log.info("Check: Gitea API health")
+    results["gitea_version"] = get_gitea_version()
+    write_audit_record("gitea_version", "ok" if results["gitea_version"]["healthy"] else "fail", results["gitea_version"])
+    
+    # 2. API baseline
+    log.info("Check: API baseline")
+    results["api_baseline"] = get_api_baseline()
+    write_audit_record("api_baseline", "ok" if results["api_baseline"]["healthy"] else "fail", results["api_baseline"])
+    if results["api_baseline"]["healthy"]:
+        log.info(f"  Baseline: {results['api_baseline'].get('baseline')}")
+    
+    # 3. Bare repo branch sync
+    log.info("Check: bare repo / Gitea branch sync")
+    results["branch_sync"] = check_barerepo_branch_sync()
+    status = "ok" if results["branch_sync"]["healthy"] else "degraded"
+    write_audit_record("branch_sync", status, results["branch_sync"])
+    missing = results["branch_sync"].get("missing_in_gitea", [])
+    if missing:
+        log.warning(f"  {len(missing)} branches missing in Gitea DB")
+        for b in missing:
+            log.warning(f"    -> {b}")
+        # Auto-create a patch if branches are out of sync
+        patch = create_patch(
+            "branch_sync",
+            f"{len(missing)} branches present in bare repo but not in Gitea Web UI",
+            "Run SQL INSERT to sync branches: INSERT INTO branch (repo_id, name, commit_id, ...) "
+            "for each missing branch. Verify with bare repo SHA.",
+            ["runtime/gitea-bin/data/... (SQL via gitea CLI or direct DB)"]
+        )
+        patches_created.append(patch)
+    
+    # 4. Stuck commands
+    log.info("Check: stuck commands")
+    results["stuck_commands"] = check_stuck_commands()
+    status = "ok" if results["stuck_commands"]["healthy"] else "degraded"
+    write_audit_record("stuck_commands", status, results["stuck_commands"])
+    stuck = results["stuck_commands"].get("stuck_commands", [])
+    if stuck:
+        log.warning(f"  {len(stuck)} commands stuck in running/")
+        for c in stuck:
+            log.warning(f"    -> {c['name']} (age: {c['age_minutes']}m)")
+    
+    # Update state
+    state["checks"] = {
+        "last_run": time.strftime("%Y-%m-%dT%H:%M:%S%z"),
+        "results": {k: {"healthy": v.get("healthy"), "error": v.get("error")} for k, v in results.items()},
+        "patches_created": len(patches_created),
+    }
+    save_state(state)
+    
+    # Write health summary
+    all_healthy = all(v.get("healthy", False) for v in results.values())
+    summary = {
+        "timestamp": time.strftime("%Y-%m-%dT%H:%M:%S%z"),
+        "controller_version": CONTROLLER_VERSION,
+        "all_healthy": all_healthy,
+        "checks": {k: v.get("healthy") for k, v in results.items()},
+        "patches_pending": len(patches_created),
+    }
+    HEALTH_FILE.parent.mkdir(parents=True, exist_ok=True)
+    HEALTH_FILE.write_text(json.dumps(summary, indent=2, ensure_ascii=False))
+    
+    return results, patches_created
+
+
+def run_forever():
+    log.info(f"Autonomous Controller v{CONTROLLER_VERSION} started")
+    log.info(f"Poll interval: {POLL_INTERVAL}s")
+    log.info(f"Workspace: {WORKSPACE}")
+    log.info(f"Gitea: {GITEA_URL}, API: {API_URL}")
+    
+    while True:
+        try:
+            log.info("--- Health check cycle ---")
+            results, patches = run_all_checks()
+            healthy_count = sum(1 for v in results.values() if v.get("healthy"))
+            total = len(results)
+            log.info(f"Results: {healthy_count}/{total} healthy")
+            if patches:
+                log.warning(f"Created {len(patches)} patch(es) for owner review")
+            else:
+                log.info("No issues detected requiring patches")
+        except Exception as e:
+            log.exception(f"Check cycle failed: {e}")
+        
+        time.sleep(POLL_INTERVAL)
+
+
+def run_once():
+    log.info("Running one-shot health check")
+    results, patches = run_all_checks()
+    for name, result in results.items():
+        status = "OK" if result.get("healthy") else "FAIL"
+        log.info(f"  [{status}] {name}")
+        if not result.get("healthy") and result.get("error"):
+            log.info(f"         error: {result['error'][:200]}")
+    if patches:
+        log.info(f"Patches created: {len(patches)}")
+    else:
+        log.info("No patches created")
+
+
+if __name__ == "__main__":
+    if len(sys.argv) > 1 and sys.argv[1] == "--once":
+        run_once()
+    else:
+        run_forever()
@@ -0,0 +1,2 @@
+# AIOS Agent State
+Agent state and lock files for AIOS Git Agent execution.
@@ -0,0 +1,12 @@
+{
+  "ts": "2026-07-09T02:08:34+0800",
+  "ok": false,
+  "checks": {
+    "gitea": false,
+    "api": true,
+    "branch_sync": false,
+    "stuck": true
+  },
+  "plans": 1,
+  "applied": 0
+}
@@ -0,0 +1,11 @@
+{
+ "ts": "2026-07-09T00:00:00+09:00",
+ "ok": true,
+ "checks": {
+ "git_object_permission": true,
+ "api": true,
+ "branch_sync": true,
+ "stale_runtime_state_cleaned": true
+ },
+ "note": "Updated during PR #2 cleanup before merge"
+}
@@ -0,0 +1,15 @@
+{
+  "processed": {
+    "CMD-20260708-fix-git-actions-write-and-enable-audit-plane.md": {
+      "hash": "9442e78e3deb35e03227016b2265d34feb4640f8efdcb07baffe06d52ef947c8",
+      "status": "running",
+      "discovered_at": "2026-07-09T00:12:52+0800"
+    },
+    "CMD-20260709-enable-autonomous-controller.md": {
+      "hash": "2825811706389130b2f531a2d3814b800b1181598f871b496e5a05fa13adbbec",
+      "status": "running",
+      "discovered_at": "2026-07-08T19:01:45+0800"
+    }
+  },
+  "last_scan": "2026-07-09T10:42:52+0800"
+}
@@ -1,22 +1,12 @@
 {
-  "version": "V1R1C0",
-  "architecture_version": "V1",
-  "release_version": "R1",
-  "change_version": "C0",
-  "components": [
-    "webops",
-    "owner-admin-user",
-    "npm",
-    "casdoor",
-    "openclaw",
-    "hermes",
-    "one-api",
-    "gitops-control",
-    "baseline-registry",
-    "custom-llm-registry",
-    "runtime-selector"
-  ],
-  "entry": "https://cnnetmkt.top",
-  "proxy": "npm",
-  "runtime_sync": "openclaw"
+  "active_version": "V1R2C3",
+  "active_tag": "V1R2C3",
+  "previous_version": "V1R2C2",
+  "rollback_version": "V1R2C2",
+  "mode": "gitops-online",
+  "update_mode": "git_patch_only",
+  "zip_update_deprecated": true,
+  "chatgpt_access": true,
+  "codex_access": true,
+  "updated_at": "2026-07-08T00:00:00Z"
 }
@@ -0,0 +1,41 @@
+# AIOS Agent Execution Audit Plane
+
+## Purpose
+Record every agent-initiated action on the AIOS GitOps runtime: what, who, when, why, result.
+
+## Audit Lifecycle
+
+1. **Pre-flight**: Record intent before execution
+2. **In-flight**: Capture command, input, git state, branch
+3. **Post-flight**: Record result, outputs, errors, duration
+4. **Archive**: Immutable commit to git as structured JSON logs
+
+## Audit Schema (v1)
+
+```json
+{
+  "audit_id": "<uuid or hash>",
+  "event_type": "command_execution | command_completed | command_failed | baseline_promotion | tag_creation | state_change",
+  "command_ref": "CMD-<date>-<description>",
+  "agent": "openclaw | hermes | codex",
+  "user": "gitops-write-bot",
+  "timestamp": "<ISO-8601>",
+  "git": {
+    "branch": "main",
+    "sha_head": "<sha>",
+    "baseline": "V1R3C1",
+    "dirty": false
+  },
+  "input": {},
+  "output": {},
+  "duration_ms": 0,
+  "result": "success | failure | pending",
+  "error": null
+}
+```
+
+## Storage
+- `audit/agent-runs/` — per-agent JSONL
+- `audit/command-history/` — per-command aggregated
+- `audit/version-history/` — per-baseline audit trail
+- All audit files committed to git (immutable)
@@ -0,0 +1,16 @@
+{
+  "agent_run_id": "openclaw-1783508505",
+  "command": "CMD-20260709-enable-autonomous-controller.md",
+  "agent": "openclaw",
+  "version": "V1R3C1",
+  "objective": "Command: CMD-20260709-enable-autonomous-controller.md",
+  "start_time": "2026-07-08T19:01:45+0800",
+  "end_time": null,
+  "interaction_count": 0,
+  "tool_call_count": 0,
+  "changed_files": [],
+  "commit_sha": null,
+  "test_result": null,
+  "final_result": "running",
+  "evidence_path": "evidence/CMD-20260709-enable-autonomous-controller/"
+}
--- a/Show More
+++ b/Show More
				`@@ -0,0 +1 @@`
				`{"owner_approved": true, "version": "V1R4C0-test"}`