docs(owner-gate): document owner maintenance gate

feat(owner-gate): add allowlist-only maintenance runner
chore(owner-gate): request memory index rebuild via command key
2026-07-10 18:33:05 +08:00 · 2026-07-10 18:32:59 +08:00 · 2026-07-10 18:32:48 +08:00 · 2026-07-10 18:32:43 +08:00 · 2026-07-10 18:32:35 +08:00
47 changed files with 320 additions and 6639 deletions
@@ -1,156 +0,0 @@
 # 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.
@@ -1,267 +0,0 @@
 # 执行 GitOps 独立 Worktree 与统一运行身份迁移
 ```yaml
 task_id: TASK-GITOPS-ISOLATED-WORKTREE-MIGRATION-20260714T233742UTC8
 created_at: 2026-07-14T23:37:42+08:00
 timezone: Asia/Shanghai
 owner_approved: true
 user_action_required: false
 risk: high
 production_business_change: false
 ```
 ## 目标
 根治共享 worktree 与多身份写入导致的 ACL、checkout 和未跟踪文件覆盖问题。
 实施：
 1. 统一 GitOps/OpenClaw 写入身份为 `aiman:gitops`；
 2. 统一 `UMask=0002`；
 3. 为 GitOps 根目录设置 setgid 与默认 ACL；
 4. 建立 bare repo 和每任务独立 worktree；
 5. 启用任务启动前自检；
 6. 验证 OpenClaw 与 ChatGPT 并发任务互不干扰；
 7. 保留原共享 workspace 作为只读回滚源，禁止智能体继续在其中切换分支。
 ## 强制通知
 收到任务后立即微信通知 Owner：
 ```text
 【任务已接收】GitOps独立worktree与统一运行身份迁移开始。将暂停写入服务、建立独立任务工作树并验证并发执行；生产业务版本不变。
 ```
 任务结束后无论 PASS、BLOCKED 或 ROLLED_BACK，立即微信汇报结果并记录 message_id。
 ## 1. 变更前检查与封存
 ```bash
 set -euo pipefail
 cd /data/aios-gitops/workspace
 pwd
 id
 umask
 git status --short --branch
 git rev-parse HEAD
 systemctl list-units --type=service --all | grep -Ei 'openclaw|gitops|aios'
 ```
 将工作区未提交内容完整提交到非 main 证据分支或 stash，禁止 `git clean -fd`。
 记录当前运行服务、用户、组、umask、PID、工作区 HEAD 和未跟踪文件清单。
 ## 2. 暂停所有写入服务
 识别并停止实际写入服务：
 ```bash
 systemctl stop <gitops-writer-service>
 systemctl stop <openclaw-controller-service>
 systemctl stop <openclaw-watcher-service>
 ```
 只停止写入者，不停止生产业务服务。记录每条命令和退出码。
 ## 3. 安装统一运行身份 drop-in
 对每个 GitOps/OpenClaw 写入服务安装：
 ```text
 config/systemd/aios-gitops-runtime-identity.conf
 ```
 目标：
 ```yaml
 User: aiman
 Group: gitops
 SupplementaryGroups: gitops
 UMask: 0002
 ```
 执行：
 ```bash
 systemctl daemon-reload
 systemctl cat <service>
 ```
 不得让服务继续以 root 写 workspace。
 ## 4. 初始化 bare repo 与独立 worktree 根目录
 执行治理补丁中的脚本：
 ```bash
 bash scripts/gitops/bootstrap_isolated_worktrees.sh
 ```
 验收：
 ```yaml
 bare_repo: /data/aios-gitops/repo.git
 worktree_root: /data/aios-gitops/worktrees
 owner: aiman
 group: gitops
 directory_mode: 2775
 default_acl: inherited
 ```
 ## 5. 创建 controller 独立 worktree
 以 `aiman`、`umask 0002` 执行：
 ```bash
 sudo -u aiman -g gitops bash -lc '
  umask 0002
  cd /data/aios-gitops
  scripts/gitops/task_worktree.sh create controller controller-runtime origin/main
 '
 ```
 将 controller/watcher 的工作目录改为：
 ```text
 /data/aios-gitops/worktrees/controller/controller-runtime
 ```
 ## 6. 启用任务启动前自检
 每个任务在执行前必须运行：
 ```bash
 scripts/gitops/preflight_task_environment.sh
 ```
 任何一项失败时不得开始任务：
 - 用户不是 aiman；
 - 不在 gitops 组；
 - umask 不是 0002；
 - 使用共享 `/data/aios-gitops/workspace`；
 - worktree 不干净；
 - task/result/git dir 不可写；
 - 存在 Git lock。
 ## 7. 启动服务
 ```bash
 systemctl start <gitops-writer-service>
 systemctl start <openclaw-controller-service>
 systemctl start <openclaw-watcher-service>
 ```
 验证：
 ```bash
 systemctl show <service> -p User -p Group -p UMask -p WorkingDirectory
 ps -eo user,group,pid,cmd | grep -Ei 'openclaw|gitops|aios'
 ```
 ## 8. 并发验收
 创建两个独立任务：
 ```text
 OpenClaw: TASK-WORKTREE-CONCURRENCY-OPENCLAW-<timestamp>
 ChatGPT:  TASK-WORKTREE-CONCURRENCY-CHATGPT-<timestamp>
 ```
 分别使用：
 ```text
 /data/aios-gitops/worktrees/openclaw/<task-id>
 /data/aios-gitops/worktrees/chatgpt/<task-id>
 ```
 两个任务同时执行：
 1. 写各自 `.agent/tasks/<agent>/`；
 2. 写各自 `.agent/results/<agent>/`；
 3. 创建深层子目录，验证默认 ACL；
 4. commit；
 5. push；
 6. 不切换共享 workspace 分支。
 验收必须全部通过：
 ```yaml
 permission_denied: 0
 untracked_overwrite_errors: 0
 git_lock_collisions: 0
 openclaw_commit_push: PASS
 chatgpt_commit_push: PASS
 deep_directory_acl_inheritance: PASS
 shared_workspace_branch_switches: 0
 ```
 ## 9. 回滚
 若迁移失败：
 1. 停止新服务；
 2. 保留失败 worktree 供审计；
 3. 恢复 systemd drop-in 备份；
 4. `daemon-reload`；
 5. 恢复原服务状态；
 6. 不删除 `/data/aios-gitops/workspace`；
 7. 生产业务保持不变。
 ## 10. 结果文件
 提交：
 ```text
 .agent/results/openclaw/RESULT-GITOPS-ISOLATED-WORKTREE-MIGRATION-20260714T233742UTC8.md
 ```
 首段必须包含：
 ```yaml
 status: PASS | BLOCKED | ROLLED_BACK
 task_id: TASK-GITOPS-ISOLATED-WORKTREE-MIGRATION-20260714T233742UTC8
 created_at: 2026-07-14T23:37:42+08:00
 timezone: Asia/Shanghai
 user_action_required: false
 runtime_identity:
  user:
  group:
  umask:
 bare_repo:
 worktree_root:
 controller_worktree:
 services:
 default_acl:
 preflight:
 concurrency_test:
  openclaw_branch:
  openclaw_commit:
  chatgpt_branch:
  chatgpt_commit:
  permission_denied_count:
  untracked_overwrite_count:
  git_lock_collision_count:
 owner_notifications:
  received_message_id:
  terminal_message_id:
 rollback:
 remaining_risks:
 next_step:
 ```
 提交并推送到独立执行分支。禁止直接推 main。
 ## 完成微信模板
 PASS：
 ```text
 【任务完成】GitOps独立worktree迁移已通过。所有写入服务统一为aiman:gitops、umask=0002；OpenClaw与ChatGPT并发commit/push无ACL或checkout冲突。生产业务版本未变化。
 ```
 阻塞：
 ```text
 【任务阻塞】GitOps独立worktree迁移未通过。原因=<具体服务/权限/并发验收>；已保留原workspace和回滚能力，生产业务未变化。
 ```
@@ -1 +1 @@
-V1R4C6
+V1R4C3
@@ -1,13 +1,8 @@
 {
-    "current_baseline": "V1R4C6",
+ "current_baseline": "V1R4C3",
-    "previous_baseline": "V1R4C5",
+ "previous_baseline": "V1R3C1",
-    "promoted_at": "2026-07-11T21:00:00+08:00",
+ "promoted_at": "2026-07-09T17:44:57+08:00",
-    "git": "e204a90398bf646d46245b3088226749fa056edd",
+ "git": "f5435e308794c737f3e39de3f6b79807486943c3",
-    "promoted_by": "openclaw/admin",
+ "promoted_by": "openclaw",
-    "promotion_reason": "V1R4C6 admin search agent integration, search gateway, agent platform, VERSION bump, merge tests deploy pipeline — not yet deployed to production",
+ "promotion_reason": "V1R4C3 autonomous operations, search gateway, task center, owner gate, and blocker self-optimization baseline promotion"
-    "constraints": {
+}
        "memory_search_enabled": false,
        "force_push_allowed": false,
        "gitea_db_mutation_requires_owner_approval": true
    }
 }
@@ -0,0 +1,48 @@
 cmd_id: CMD-20260710-memory-index-rebuild
 operation: memory.index.rebuild
 risk: medium
 permission: OWNER_NON_DANGEROUS_ALL
 status: pending_owner_gate
 reason: Memory search is disabled because existing embedding index metadata does not match the current embedding provider configuration.
 requested_by: codex-chartgpt-bridge
 requested_at: "2026-07-10T08:00:00+08:00"
 command_key: openclaw.memory.index.force
 args:
  force: true
 safety:
  destructive: false
  arbitrary_shell: false
  credential_mutation: false
  provider_mutation: false
  external_network_write: false
  production_data_delete: false
  baseline_or_tag_mutation: false
  main_branch_direct_write: false
 owner_gate:
  required: true
  approval_model: protected_branch_merge_or_signed_owner_commit
  approved_by: null
  approved_at: null
  approval_commit: null
 execution:
  runner: owner-maintenance-runner
  lock_key: memory-index-rebuild
  timeout_seconds: 1800
  executed_by: null
  executed_at: null
  exit_code: null
 acceptance:
  commands:
    - openclaw memory search "V1R4C3"
    - openclaw memory search "Gitea Branch Sync"
    - openclaw memory search "Autonomous Controller"
  expected:
    - no_embedding_metadata_mismatch
    - memory_search_returns_results
    - memory_index_uses_current_embedding_provider
 success_signals:
  - AIOS_OWNER_MAINTENANCE_MEMORY_INDEX_REBUILT
  - AIOS_MEMORY_SEARCH_RESTORED
 rollback:
  destructive: false
  strategy: rebuild_previous_index_if_snapshot_exists
@@ -1,59 +0,0 @@
 version: 1
 policy_id: gitops-isolated-worktree-governance-20260714T232948UTC8
 runtime_identity:
  user: aiman
  group: gitops
  umask: "0002"
  prohibit_root_workspace_writes: true
 repository:
  source_worktree: /data/aios-gitops/workspace
  bare_repo: /data/aios-gitops/repo.git
  worktree_root: /data/aios-gitops/worktrees
  remote: origin
 worktrees:
  controller: /data/aios-gitops/worktrees/controller
  task_pattern: /data/aios-gitops/worktrees/{agent}/{task_id}
  allowed_agents:
    - openclaw
    - chatgpt
    - controller
  require_clean_start: true
  remove_after_successful_push: true
  retain_on_failure: true
 permissions:
  owner: aiman
  group: gitops
  directory_mode: "2775"
  file_mode: "0664"
  default_acl:
    user:aiman: rwx
    group:gitops: rwx
  apply_to:
    - /data/aios-gitops
    - /data/aios-gitops/worktrees
 preflight:
  required_user: aiman
  required_group: gitops
  required_umask: "0002"
  require_writable_git_dir: true
  require_writable_task_dir: true
  require_writable_result_dir: true
  reject_shared_source_worktree: true
  reject_existing_git_lock: true
 notifications:
  owner_on_received: required
  owner_on_terminal: required
  preferred_channel: weixin
  redact_secrets: true
 rollback:
  preserve_source_worktree: true
  remove_created_worktrees: true
  restore_service_dropins: true
  restore_service_state: true
@@ -1,14 +0,0 @@
 # Install as a systemd drop-in for GitOps/OpenClaw writer services.
 # Example: /etc/systemd/system/<service>.service.d/10-runtime-identity.conf
 [Service]
 User=aiman
 Group=gitops
 SupplementaryGroups=gitops
 UMask=0002
 Environment=AIOS_GITOPS_ROOT=/data/aios-gitops
 Environment=AIOS_BARE_REPO=/data/aios-gitops/repo.git
 Environment=AIOS_WORKTREE_ROOT=/data/aios-gitops/worktrees
 Environment=AIOS_RUNTIME_USER=aiman
 Environment=AIOS_RUNTIME_GROUP=gitops
 Environment=AIOS_RUNTIME_UMASK=0002
@@ -1,7 +0,0 @@
 # V1R4C6 Backup Result
 - version: V1R4C6
 - status: NOT_STARTED
 - reason: No confirmed production Stack, persistent volume, database migration target, or V1R4C5 image digest is bound to this source release. A backup cannot be claimed or executed safely without an identified target datastore.
 - prohibited_actions_not_performed: production database deletion; persistent-volume deletion; docker system prune --volumes; direct copy of live database files.
 - required_before_resume: confirmed Stack identity and its documented backup/restore command.
@@ -1,14 +0,0 @@
 # V1R4C6 Deployment Result
 - version: V1R4C6
 - source_commit: e204a90398bf646d46245b3088226749fa056edd
 - tag_commit: d8cf6058e99f5c7f2dfd2b45a406c0a7edf955b2
 - image_repository: UNRESOLVED
 - image_digest: UNRESOLVED
 - previous_image_digest: UNRESOLVED
 - portainer_environment: UNRESOLVED
 - portainer_stack: UNRESOLVED
 - deployment_action: NOT_EXECUTED
 - reason: No source-to-image-to-Portainer-Stack mapping exists. Existing Portainer containers and known Docker workloads do not expose a V1R4C6/e204 deployment target or Portainer stack labels. The active GitOps API is a direct systemd Python process, not a Portainer-managed Docker Stack.
 - rollback_status: NOT_APPLICABLE_NO_CHANGE
 - final_status: BLOCKED_MISSING_CONFIRMED_PRODUCTION_STACK
@@ -1,15 +0,0 @@
 # V1R4C6 Production Preflight Result
 - version: V1R4C6
 - started_at_utc: 2026-07-11T16:57:04.0596039Z
 - source_tag_object: 356a8a8c1860b94b7d61a456c29228c88e57d963
 - source_tag_commit: d8cf6058e99f5c7f2dfd2b45a406c0a7edf955b2
 - source_commit: e204a90398bf646d46245b3088226749fa056edd
 - tag_to_source_relation: tag commit is direct parent of source commit (verified)
 - production_host_identifier: cn-network-mkt (hw / 1.95.164.183)
 - host_preflight: Docker 29.5.3; root filesystem 29 percent used; memory available approximately 56 GiB
 - GitOps_runtime: active at /data/aios-gitops/workspace via aios-gitops-model-api.service
 - runtime_base: current_baseline V1R4C6; previous_baseline V1R4C5; git e204a90398bf646d46245b3088226749fa056edd; service status ok
 - Portainer discovery: Portainer instances are reachable on ubuntu-admin and ubuntu-aihost, but no confirmed Portainer Environment/Stack maps to V1R4C6 or e204a90398bf646d46245b3088226749fa056edd.
 - source_build_contract: target source contains only deploy/gitea/docker-compose.yml and no application Dockerfile/Containerfile or V1R4C6 image repository declaration.
 - status: BLOCKED_MISSING_CONFIRMED_PRODUCTION_STACK
@@ -1,7 +0,0 @@
 # V1R4C6 Rollback Result
 - version: V1R4C6
 - rollback_required: No
 - rollback_executed: No
 - reason: No container, Stack, database, image, or production configuration was changed during this run.
 - captured_v1r4c5_digest: UNRESOLVED because no matching current production Stack was confirmed.
@@ -1,11 +0,0 @@
 # V1R4C6 Verification Result
 - version: V1R4C6
 - healthcheck_result: GitOps adapter /health = 200 and /api/base = 200; this verifies the control-plane API only, not a V1R4C6 production deployment.
 - gate_a_result: NOT_RUN (production admin workbench target and authenticated test path unresolved)
 - gate_b_result: NOT_RUN (production Search Gateway target/provider binding unresolved)
 - gate_c_result: NOT_RUN (production Agent Platform target/runtime request path unresolved)
 - runtime_acceptance_result: NOT_RUN (0/8; cannot reuse non-production results)
 - critical_control_plane_finding: gitea-branch-consistency-sentinel is failed with five Git-ref-to-Gitea-metadata drift entries and autonomous safe mode emitted. This is a hard safety gate failure.
 - no_unresolved_critical_errors: false
 - final_status: BLOCKED_MISSING_CONFIRMED_PRODUCTION_STACK
@@ -1,113 +0,0 @@
 # 微信文本-图片缓存隔离事故报告
 ## 结论
 本次事故已经按最新备份链路回滚并完成第二轮评审修复。`3d9782fc346d1e7defde090ab8637e62ab28ab8a` 进一步修复了英文诊断文本和未确认引用文本误入图片路径的问题，`706c0b0` 则是后续最小修复提交。当前文档只记录证据、部署和验收边界，不把“真实 owner inbound 已完成”写成结论。
 ## 现象
 - 纯文本消息被误答成旧图片回复，属于 stale-image reply on pure text。
 - 现场证据显示 `types=1` 且 `hasMedia=false`，但后续又进入 `message_type=image`、`image_count=0`、`implicit_image_message`，最终落到 `cached_image_expired`。
 - 这说明“诊断文本”被误判成图片意图后，又继承了 sender 级旧图缓存。
 ## 根因
 - `fixes/patch_openclaw_weixin_hermes.py:768-839` 把文本意图判断和缓存回退放在同一条链上。
 - `looksLikeHermesImageFollowup()` 会把含图片关键字的诊断文本推进 image intent。
 - `resolveHermesFollowupImageContext()` 随后会读取 sender 级缓存；一旦缓存命中失败或过期，就会落到 `cached_image_expired`，从而让纯文本沾上旧图上下文。
 - `handleHermesImplicitImage()` 继续把消息送入 `handleHermes(..., "implicit_image_message")`，所以误判会沿着同一条图片处理通道传播。
 ## 护栏不变量
 - 图片处理上下文必须来自当前消息图片、已确认的引用图片，或精确匹配的缓存图片 ID；缺少这三类证据时，纯文本不能进入 `handleHermesImplicitImage()`。
 - 未确认的引用文本不能继承 sender 级旧图缓存，必须返回 `referenced_image_not_confirmed`。
 - 只有精确缓存图片 ID 命中且缓存仍有效时，才允许走 `resolveHermesCachedImageEntry()` 的 `cached_image_matched` 路径。
 - `/hermes` 这类 slash 路径保持原有分流，slash/channel 面不变。
 - 这次修复的守卫目标，是把中英文诊断文本、未确认引用文本与图片缓存回退拆开，避免再把普通文本提升成图片意图。
 ## TDD 与幂等性
 - `66d637fc0364a617c7b9fd8cd9baa68fb0368675`（`fix(weixin): isolate text from stale image cache`）引入了隔离修复和 `tests/test_weixin_image_followup_intent.py`。
 - `2de29c59bb946cc3db3e14e8a06270391ede8401`（`fix(weixin): add process-only runtime patch mode`）继续收紧 process-only 运行模式，保持回放幂等。
 - `3d9782fc346d1e7defde090ab8637e62ab28ab8a`（`fix(weixin): require confirmed image reference context`）补齐英文诊断文本和未确认引用文本用例，并把图片引用收紧到已确认上下文。
 - 用户实测的 TDD 路径为 RED `8 PASS / 1 FAIL`，随后 GREEN `9 PASS`。
 - 最小修复提交为 `706c0b0`，对应 quoted pure text 的最后一层收敛修复。
 - 第二轮回归结果为 `8 PASS`；node check 同样通过，重复应用补丁保持幂等。
 - 生产 npm 分发侧本身缺少 `tsconfig/tests`，baseline 与 patched 共享这一限制，所以这里不把分发目录的缺口当成回归点；验收以 node check 和 8 tests 为准。
 ## 用户实测补充
 - 普通文本两项为 PASS。
 - quoted pure text 为 FAIL，日志证据包含 `ref_msg.message_item.type=0/NONE`、quoted id、`implicit_image_message`、`referenced_image_unresolved`。
 - 这组证据说明 quoted pure text 没有被当成真正的图片 follow-up，且没有越过引用确认边界。
 ## 回滚与备份
 - 回滚按最新备份执行，基线指向 `/data/aios-gitops/runtime/latest-backup-dir.txt` 所示的最新备份目录。
 - 备份路径链路按供给证据记录为 `T064506Z`、`T074212Z`、`T081658Z`。
 - 第一轮恢复后的部署过程哈希记录为 `e9625d12` 和 `c00c58f8`。
 - v2 部署备份为 `/var/backups/aios/weixin-text-cache-isolation-20260711T094908Z`。
 - v2 部署文件哈希为 TS `024c6498284d1871b048f6157d5d3dbd4d34e54b53a05afd744b00458ceeb79f`、JS `d73b15d6752056192a1d1939b04107334419785b41175b39f1cbc1132ac65c47`；slash/channel 口径保持不变。
 - 首次部署因为 `-- No entries --` 门禁误判而回滚，随后纠正部署在线，PID 为 `2496603`。
 - 在线修复后的 hashes 为 `86dfe...` / `a9b854...`，`channel 4` accounts healthy，cache unchanged。
 ## 主线合并与生产发布
 - 主线 fast-forward source rev 为 `9269e819972c2e709e0c27bed7eab590da6961c9`，本次不 tag、不 baseline。
 - 生产发布结论为 PASS，release dir 为 `/var/backups/aios/weixin-mainline-release-20260711T181842Z`。
 - old PID 为 `2655072`，new PID 为 `2656106`。
 - TS hash 为 `86dfe4543c4414bc86778a4b218c244a742a139c66e3c9f19cf6f0de871234c3`，JS hash 为 `a9b854015108020458ce6f31b9d9187eff82e170c4c214259b0afb9be6a7336f`。
 - `18789` 监听，`channel 4` accounts healthy，`error_count=0`，cache unchanged。
 - 一次参数门禁失败已回滚，且不影响最终发布。
 - owner quoted-text live acceptance 仍为 PENDING。
 ## 运行态验证
 - 第一轮 gateway PID 由 `4025932` 变为 `1723576`；v2 部署后由 `1723576` 变为 `1876713`。
 - v2 现场/供给证据为 `active`、监听 `18789`、`error_count=0`、`channel gate pass`、`cache preserved`。
 - 最终 read-only gate 通过，说明部署后运行态未出现新的插件错误，channel 状态和缓存均保持预期。
 - quoted pure text 的第二次 owner live acceptance 仍为 PENDING。
 ## 未完成的真实验收
 - owner 纯文本
 - diagnostic text
 - genuine image-followup message
 - quoted pure text 的第二次 owner live acceptance
 以上四类消息仍需要由 owner 真实 inbound 到达，才能把这次“文本隔离”结论升格为完整的线上接受；本报告明确保留该项为 pending，不把它们写成已经完成。
 ## 引用文字语义修复与发布（2026-07-12）
 - 后续 owner 实测证明图片误路由已消失，但模型在没有被引正文时根据 `chat_id/message_id/timestamp` 编造解释，因此图片隔离 PASS、语义验收 FAIL。
 - 现场 payload 的 `ref_msg.message_item.type=0/NONE` 只有 quoted ID 与时间字段，没有 `text_item`；session transcript 又没有原始 Weixin 数字 message ID，不能精确回表。
 - 出站 `clientId` 与引用的 19 位数字 `msg_id` 格式不同，发送 API 也不返回服务端 message ID。本次明确禁止 recent-message、时间窗口、session 顺序和近似 ID 猜测。
 - `1136b4186d7578cd77f255297204b10152c47dab` 增加证据优先处理：payload 明确含 `text_item.text` 时才构造引用上下文；引用存在但正文缺失时，经标准 dispatcher 确定性回复“无法读取被引用文字，请复制原文后再提问。”，并在模型 dispatch 前返回。
 - TDD 先出现两项预期 RED（quoted-text resolver 不存在），最终 GREEN 为 `17 passed`；`py_compile`、`git diff --check`、`node --check` 均通过。
 - Scratch 双跑验证只改变 `src/messaging/process-message.ts` 与 `dist/src/messaging/process-message.js`，第二次应用 diff 为 0；slash/channel 文件保持不变。
 - 主线为 `1136b4186d7578cd77f255297204b10152c47dab`，生产发布目录为 `/var/backups/aios/weixin-quoted-text-failclosed-release-20260712T054451Z`。
 - 发布后 PID 从 `2656106` 变为 `3670157`；TS hash 为 `1a73e735e9c8fb40a1f82d2a66bc022396c8105ed989aac95e11b66326e9fdf1`，JS hash 为 `1c76bf6ec3b45021e68b932cbf52222ba2130a867b95e41d7a91f261b991a304`。
 - `18789`、4 个 Weixin accounts、post-start error 0 和图片 cache unchanged 均通过。新的 owner live acceptance 仍为 PENDING。
 ## Owner 引用文字 Fail-Closed 验收（2026-07-12）
 - Owner 引用一条微信未提供正文的文字消息并发送“是什么意思”，收到固定回复“无法读取被引用文字，请复制原文后再提问。”。
 - 日志窗口为 `2026-07-12 14:06:23.424-14:06:23.893 +08:00`：入站 `types=1`、`bodyLen=6`、`hasMedia=false`，出站 `textLen=20` 且发送成功。
 - 同一窗口 `model-fetch=0`、`implicit_image_message=0`、Hermes route=0，`quoted_text_unavailable=1` 且 `quoted_message_id=present`。
 - 该证据确认无正文引用不会进入模型猜测或图片缓存路径。Owner quoted-text no-body live acceptance 为 **PASS**。
 - 微信 payload 明确携带 `text_item.text` 时的 grounded explanation 和 genuine image follow-up 仍作为独立兼容性验收项保留。
 ## 可见引用摘要解析修复（2026-07-12）
 - Owner 进一步确认微信客户端能够显示被引用正文，因此上一节的“未提供正文”只适用于旧 resolver 的观察结果，不代表协议 payload 没有可用文本。
 - 根因是插件 `inbound.ts` 已将协议字段 `ref_msg.title` 定义为引用摘要并拼入 `ctx.Body`，而 `extractHermesReferencedText()` 只读取 `ref_msg.message_item.text_item.text`，导致客户端可见摘要被错误丢弃。
 - `13c2a182860131abcad0f5a70ae8263477578588` 在真实 `text_item.text` 缺失时，仅从 reference envelope 读取 `ref_msg.title`；不会读取 `message_item.title` 或任意嵌套通用 `title`，也不改变图片、slash 或 channel 路由。
 - TDD 先得到 `18 passed / 1 failed`，失败项为可见 `ref_msg.title` 仍落到 `quoted_text_unavailable`；修复后为 `19 passed`，`py_compile`、`node --check` 与 `git diff --check` 通过。
 - 生产发布目录为 `/var/backups/aios/weixin-quoted-text-title-resolution-20260712T155112+0800`，PID 从 `3859376` 变为 `3860863`，post-start error 为 0，图片 cache unchanged。
 - 部署后 TS hash 为 `8e77ce61214688a2507b07f0cf1ee85401b58198b8271bebd69efc2681872726`，JS hash 为 `663db8d30a5cd5eb80fe46cf941cce6e6f5d04cb7e4b8d9b4d0086d87faba290`。
 - 可见正文引用的 owner live acceptance 尚待下一条真实微信消息确认；预期日志必须命中 `quoted_text_resolved`，且同一窗口不得出现 `quoted_text_unavailable` 或图片路由。
 ## 安全说明
 - 本次仅更新本报告，没有修改代码、runtime、services 或 secrets。
 - 报告里没有任何凭据、密钥或原始敏感 ID。
@@ -0,0 +1,49 @@
 # Owner Maintenance Gate
 This mechanism allows owner-approved non-dangerous maintenance actions without giving agents arbitrary host CLI access.
 ## Permission tiers
 - READ_ONLY: status, health, baseline, versions, search inspection.
 - AUTO_PATCH: branch creation, repo file changes, patch commits, test/report artifacts.
 - OWNER_NON_DANGEROUS_ALL: owner may approve all non-destructive allowlisted maintenance command_keys.
 - OWNER_HIGH_RISK: main merge, baseline promotion, tag creation, credential/provider mutation, destructive cleanup, production restart, SSH/firewall changes.
 ## Execution rule
 Agents can create requests. The runner can only execute fixed allowlist mappings:
 ```text
 command_key: openclaw.memory.index.force
 argv: /usr/local/bin/openclaw memory index --force
 ```
 The runner rejects arbitrary shell strings such as `bash -c`, `sh -c`, `curl | bash`, or recursive destructive commands.
 ## Owner gate
 A request becomes executable only after protected-branch merge or signed owner commit fills:
 ```yaml
 owner_gate:
  approved_by: <owner>
  approved_at: <timestamp>
  approval_commit: <signed-owner-commit-sha>
 ```
 ## Memory index rebuild flow
 1. Detect embedding metadata mismatch.
 2. Create `commands/owner-requests/CMD-20260710-memory-index-rebuild.yaml`.
 3. Owner approves through GitOps owner gate.
 4. `runtime/owner_maintenance/allowlist_runner.py` validates safety metadata.
 5. Runner executes the fixed command vector.
 6. Runner records exit code and stdout/stderr hashes.
 7. Acceptance checks verify memory search works again.
 Expected signals:
 ```text
 AIOS_OWNER_MAINTENANCE_MEMORY_INDEX_REBUILT
 AIOS_MEMORY_SEARCH_RESTORED
 ```
@@ -1,212 +0,0 @@
 # V1R4C5 Acceptance Evidence
 Date: 2026-07-11
 Status: release candidate evidence in progress
 Current active baseline: V1R4C3
 V1R4C5 has not been promoted yet.
 ## Owner Decision
 Owner approved continued V1R4C5 work under the following constraints:
 ```text
 允许无 force-push 的分支 reconcile
 memory search 保持关闭
 先完成 SR-01/SR-02/数据隔离验收
 再进入 merge/deploy/tag owner gate
 ```
 ## Scope
 V1R4C5 release candidate scope:
 1. SR-01 Gitea reliability and branch consistency guardrails.
 2. SR-02 watcher/stuck command cleanup.
 3. User data isolation redline and enforcement gates.
 4. SR-04 admin workbench read-only visibility preparation only.
 Explicitly excluded from V1R4C5 activation:
 - memory search activation;
 - full SR-04 portal production rollout;
 - bulk merge of old test/smoke/auth branches;
 - force push reconciliation;
 - Gitea DB mutation without owner/admin approval.
 ## Runtime Acceptance Evidence
 ### Runtime status
 Observed:
 ```json
 {
  "service": "aios-gitops-model-api",
  "status": "active",
  "workspace_branch": "patch/v1r4c5-runtime-acceptance-clean-20260711-01"
 }
 ```
 ### Baseline
 Observed:
 ```json
 {
  "current_baseline": "V1R4C3",
  "previous_baseline": "V1R3C1",
  "promoted_by": "openclaw"
 }
 ```
 ### Clean createBranch acceptance
 Branch:
 ```text
 test/v1r4c5-runtime-acceptance-clean-20260711-01
 ```
 Result:
 ```text
 status: ok
 commit: 6ae7058
 sync: synced
 ```
 ### Clean createPatch acceptance
 Patch:
 ```text
 patch/v1r4c5-runtime-acceptance-clean-20260711-01
 ```
 Patch id:
 ```text
 patch-1783706123789-f5ca8b69
 ```
 Commit:
 ```text
 ba0391e
 ```
 Result:
 ```text
 synced
 ```
 ## SR-01 Evidence
 Known accepted evidence:
 - Gitea branch consistency sentinel deployed.
 - Sentinel reports:
 ```json
 {
  "status": "healthy",
  "unhealthy_checks": []
 }
 ```
 - systemd timer runs every 5 minutes.
 - service runs with:
 ```text
 --emit-safe-mode --json
 ```
 - broken symlink issue resolved by replacing the external reliability symlink with workspace-local files.
 Required before final promotion:
 - confirm latest sentinel report remains healthy after acceptance writes.
 ## SR-02 Evidence
 Known accepted evidence:
 ```text
 11 / 11 tests pass
 ```
 Fix summary:
 - stuck command detection reads parsed timestamps through `data.get('ts', 0)`;
 - `.md` fixture mtime is aligned with stored stuck timestamp via `os.utime(path, (ts, ts))`;
 - behavior is consistent across `.json` and `.md` state formats.
 Required before final promotion:
 - re-run SR-02 tests on the final integration candidate.
 ## User Data Isolation Evidence
 Implemented controls:
 - user data isolation redline;
 - memory scope guard;
 - provider gate;
 - activation gate;
 - audit service/timer;
 - CI acceptance scripts;
 - memory search activation check.
 Required safety state:
 ```text
 AIOS_MEMORY_ENABLED=false
 ```
 Required before memory/search activation:
 - isolation audit healthy;
 - no denied records;
 - no invalid records;
 - provider filters by scope before ranking;
 - provenance required;
 - global index disabled;
 - activation gate returns allowed;
 - explicit owner approval.
 V1R4C5 does not approve memory search activation.
 ## SR-04 Admin Workbench Evidence
 Current user-visible gap is confirmed:
 - administrator cannot yet see task progress;
 - administrator cannot yet see flow/process diagrams;
 - SR-04 remains activation-ready but gated.
 V1R4C5 scope allows only read-only visibility preparation and acceptance design, not full production rollout.
 Required before SR-04 deployment:
 - SR-01 green;
 - SR-02 green;
 - user data isolation gate accepted;
 - admin RBAC read-only check green.
 ## Remaining Required Gates Before Promotion
 - [ ] Latest Gitea sentinel report remains healthy.
 - [ ] SR-02 watcher tests pass on final candidate.
 - [ ] User data isolation acceptance passes.
 - [ ] Admin workbench read-only RBAC acceptance passes or remains explicitly deferred.
 - [ ] No force-push was used.
 - [ ] Memory search remains disabled.
 - [ ] Owner approves merge/deploy/tag/promotion.
 ## Promotion Rule
 V1R4C5 may not be promoted until all required gates are green or explicitly deferred with owner approval.
@@ -1,54 +0,0 @@
 # V1R4C5 Final Owner Approval
 Date: 2026-07-11
 Status: approved for final promotion
 ## Owner Decision
 Owner approved final V1R4C5 promotion.
 Memory search must remain disabled.
 ```text
 允许最终晋升
 继续保持 memory search 关闭
 ```
 ## Promotion Scope
 Approved for V1R4C5:
 1. SR-01 Gitea reliability and branch consistency guardrails.
 2. SR-02 watcher/stuck command cleanup.
 3. User data isolation redline and enforcement gates.
 4. SR-04 admin workbench read-only visibility preparation only.
 ## Explicit Non-Activation
 The following are not approved by this promotion:
 ```text
 memory search activation
 full cross-session memory recall
 global vector index
 full SR-04 production portal rollout beyond read-only/admin visibility preparation
 bulk merge of old test/smoke/auth branches
 force-push based reconciliation
 ```
 ## Required Runtime State
 Memory remains disabled:
 ```text
 AIOS_MEMORY_ENABLED=false
 ```
 If a provider or deployment attempts to enable memory/search, it must be blocked unless a future owner approval explicitly allows it after the user-data isolation activation gate returns allowed.
 ## Final Promotion Rule
 V1R4C5 may proceed to tag/promotion under owner gate.
 If any final gate fails, promotion must stop and memory/search must remain disabled.
@@ -1,107 +0,0 @@
 # V1R4C5 Release to Main Reconcile Plan
 Date: 2026-07-11
 Status: reconcile preparation
 ## Objective
 Bring the V1R4C5 release state back into `main` without rewriting history and without moving the immutable V1R4C5 tag.
 ## Source of Truth
 Release baseline:
 ```text
 V1R4C5
 ```
 Promotion commit:
 ```text
 a32c76a3c1baa49562e53fb8adce10a5ed0e6704
 ```
 Tag:
 ```text
 V1R4C5
 ```
 Tag policy:
 ```text
 immutable
 ```
 ## Reconcile Rules
 Allowed:
 - normal merge commits;
 - fast-forward where possible;
 - conflict resolution with review;
 - updating main to contain released C5 state.
 Forbidden:
 - force-push;
 - tag movement;
 - rewriting release history;
 - importing V1R4C6 feature code into this reconcile.
 ## Expected Included C5 Content
 - SR-01 Gitea reliability guardrails.
 - Gitea branch consistency sentinel.
 - SR-02 watcher state cleanup.
 - User Data Isolation gate.
 - Reliability governance documents.
 - Admin Workbench read-only preparation records.
 ## Expected Excluded Content
 Do not include:
 - V1R4C6 Admin Workbench implementation code.
 - SR-03 Search Gateway implementation.
 - Agent Platform integration.
 - Memory search activation.
 - Global vector index.
 ## Validation Before Main Merge
 Required:
 ```text
 getBaseline.current_baseline == V1R4C5
 V1R4C5 tag unchanged
 memory search disabled
 Gitea sentinel healthy
 SR-02 tests pass
 main merge does not include V1R4C6 branches
 ```
 ## Post Merge Validation
 Run:
 ```text
 getRuntimeStatus
 getBaseline
 createBranch
 createPatch
 ```
 Expected:
 ```text
 main contains V1R4C5 release state
 active baseline remains V1R4C5
 ```
 ## Owner Gate
 Normal merge to `main` may proceed after validation.
 No tag creation or baseline promotion is required because V1R4C5 is already promoted.
@@ -1,137 +0,0 @@
 # V1R4C6 Acceptance Checklist
 Date: 2026-07-11
 Base baseline: V1R4C5
 Promotion blocked until every mandatory gate is green.
 ## Global Preflight
 - [ ] `getRuntimeStatus.status == active`
 - [ ] `getBaseline.current_baseline == V1R4C5`
 - [ ] `getBaseline.git == a32c76a3c1baa49562e53fb8adce10a5ed0e6704`
 - [ ] `AIOS_GITOPS_BARE_REPO` points to `/data/aios-gitops/runtime/gitea-bin/data/repositories/aios/aios.git`
 - [ ] Gitea consistency sentinel reports `healthy`
 - [ ] `AIOS_MEMORY_ENABLED=false`
 - [ ] no force-push used
 ## Gate A — Admin Workbench / Portal AIOS
 ### Route/API
 - [ ] `/portal/me/aios/manifest` returns authenticated module manifest
 - [ ] `/portal/me/aios/dashboard` returns admin-authorized dashboard data
 - [ ] `/portal/me/aios/health` returns system health data
 - [ ] failed/empty backend data produces explicit empty/error state
 ### UI
 - [ ] admin sees task progress
 - [ ] admin sees process/flow diagram entry
 - [ ] admin sees system health
 - [ ] admin sees authorized agent observability
 - [ ] user does not see admin-only modules
 - [ ] admin does not see owner-only modules without owner role
 ### Isolation
 - [ ] no cross-user data
 - [ ] no cross-project data
 - [ ] no cross-session data
 - [ ] no cross-tenant/group data
 Gate A result: `PASS / FAIL / DEFER`
 ## Gate B — SR-03 Search Gateway
 ### Provider Binding
 - [ ] real provider configured and reachable
 - [ ] provider errors return safe non-memory fallback
 - [ ] timeout/retry/circuit-breaker behavior verified
 ### Scope and Provenance
 - [ ] scope filter occurs before ranking
 - [ ] provenance returned for accepted results
 - [ ] denied/out-of-scope content never reaches ranking or summary
 - [ ] global memory/vector index disabled
 - [ ] memory search remains disabled
 ### Audit
 - [ ] search request scope is logged without leaking content
 - [ ] provider/fallback decision is auditable
 - [ ] denied scope mismatch is auditable
 Gate B result: `PASS / FAIL / DEFER`
 ## Gate C — Agent Platform
 ### Event Bus
 - [ ] event envelope includes tenant/group, user, project, session and agent scope
 - [ ] cross-scope subscribers cannot receive events
 - [ ] failed delivery is observable and retry-bounded
 ### GPT Message Plane
 - [ ] message envelope includes scope and provenance
 - [ ] messages cannot cross user/project/session boundaries
 - [ ] denied messages do not forward content
 ### Controller Dispatch
 - [ ] dispatch stops when autonomous safe mode is active
 - [ ] consequential actions require owner gate
 - [ ] duplicate dispatch is idempotent or rejected
 - [ ] stuck state cleanup remains green
 ### Observability
 - [ ] admin can see platform status without unrestricted content access
 - [ ] failures and queue state are visible
 Gate C result: `PASS / FAIL / DEFER`
 ## Combined Runtime Acceptance
 Run after integration and deployment:
 ```text
 getRuntimeStatus
 getBaseline
 createBranch
 createPatch
 ```
 Expected before promotion:
 ```text
 runtime = active
 baseline = V1R4C5
 write branch = committed + synced
 write patch = committed + synced
 ```
 ## Rollback Validation
 - [ ] Group A can disable workbench routes/menu entries
 - [ ] Group B can disable provider binding and use safe non-memory fallback
 - [ ] Group C can disable expanded dispatch
 - [ ] all rollback modes preserve `AIOS_MEMORY_ENABLED=false`
 ## Release Decision
 V1R4C6 is tag-ready only when:
 ```text
 Global preflight PASS
 AND Gate A PASS
 AND Gate B PASS
 AND Gate C PASS
 AND combined runtime acceptance PASS
 ```
 Any failed mandatory gate blocks tag, deployment promotion, and baseline promotion.
@@ -1,51 +0,0 @@
 # V1R4C6 Acceptance Evidence
 Date: 2026-07-11
 ## Merge
 - **PR #6** (merge commit `ae49bf0fa35b0f599a96a4d12644576447f39cec`): V1R4C6 release-candidate control artifacts merged to main.
 - **Result**: PASS
  - 10 files (1440 lines) added: release docs, patch records, ops commands, version metadata
  - No V1R4C6 tag created
  - No baseline promotion
  - Memory search remains disabled
  - Force-push remains disabled
 ## Verified State (post-merge)
 | Check                    | Result |
 |--------------------------|--------|
 | main HEAD                | ae49bf0fa35b0f599a96a4d12644576447f39cec |
 | V1R4C5 tag               | Still at d457ac396147043ed5d0a3fd8b44bcb10f48bceb (not moved) |
 | V1R4C6 tag               | absent |
 | baseline/current.json    | current_baseline = V1R4C5, git = a32c76a3c1baa49562e53fb8adce10a5ed0e6704 |
 | memory_search_enabled    | false |
 | force_push_allowed       | false |
 | versions/V1R4C5.json     | exists |
 | versions/V1R4C6.json     | exists |
 ## Gates
 | Gate                         | Status  | Notes                          |
 |------------------------------|---------|--------------------------------|
 | control artifacts merge      | PASS    | PR #6 merged to main           |
 | Gate A — Admin Workbench     | NOT RUN | Engineering integration pending|
 | Gate B — SR-03 Search Gateway| NOT RUN | Engineering integration pending|
 | Gate C — Agent Platform      | NOT RUN | Engineering integration pending|
 | tests                        | NOT RUN | No test suite executed         |
 | artifact digest              | NOT CREATED | No build performed          |
 | deployment                   | NOT RUN | No deploy performed            |
 | production verification      | NOT RUN | No production verification     |
 ## Rollback
 Revert command (creates new branch and PR, non-destructive):
 ```bash
 git revert -m 1 ae49bf0fa35b0f599a96a4d12644576447f39cec
 ```
 ## Current Overall Status
 **BLOCKED / INTEGRATION PENDING** — all engineering gates remain unexecuted.
@@ -1,80 +0,0 @@
 # V1R4C6 Blocker: Baseline Visibility Discrepancy
 Date: 2026-07-11
 Status: blocking merge/deploy/tag
 ## Summary
 OpenClaw/operator reported V1R4C5 baseline promotion completed successfully, but the ChatGPT GitOps action view still reports the active baseline as V1R4C3.
 V1R4C6 must not proceed to merge/deploy/tag until this view is reconciled.
 ## Operator Reported State
 Reported by operator:
 ```text
 current_baseline = V1R4C5
 previous_baseline = V1R4C3
 promotion_git = a32c76a3c1baa49562e53fb8adce10a5ed0e6704
 promoted_by = openclaw/admin
 memory_search_enabled = false
 force_push_allowed = false
 gitea_db_mutation_requires_owner_approval = true
 ```
 ## ChatGPT GitOps Action Observed State
 Observed through `getBaseline`:
 ```text
 current_baseline = V1R4C3
 previous_baseline = V1R3C1
 git = f5435e308794c737f3e39de3f6b79807486943c3
 promoted_by = openclaw
 ```
 Observed through `listVersions`:
 ```text
 V1R4C5 tag exists
 versions_dir does not list canonical V1R4C5.json
 workspace_branch = release/v1r4c6-admin-search-agent-integration-20260711
 ```
 ## Risk
 If V1R4C6 proceeds while different controllers disagree on the active baseline, the release may be built against the wrong base and post-deployment verification may be invalid.
 ## Required Reconciliation
 Before continuing V1R4C6:
 1. Confirm the GitOps API used by ChatGPT reads the same `baseline/current.json` updated by OpenClaw.
 2. Confirm the API service has reloaded or refreshed baseline state.
 3. Confirm canonical `versions/V1R4C5.json` is visible to the action runtime or intentionally explain why it is not required.
 4. Re-run:
 ```text
 getBaseline
 listVersions
 getRuntimeStatus
 ```
 Required result:
 ```text
 getBaseline.current_baseline = V1R4C5
 listVersions.tags contains V1R4C5
 memory search remains disabled
 ```
 ## Hard Stop
 Do not create V1R4C6 tag.
 Do not promote V1R4C6 baseline.
 Do not deploy V1R4C6.
 Do not enable memory search.
 Proceed only after baseline visibility is reconciled.
@@ -1,83 +0,0 @@
 # V1R4C6 Context Isolation Recovery and Resume Decision
 Date: 2026-07-11
 Status: approved to resume merge/deploy/verification; stop before tag/promotion
 ## Recovery Evidence
 The unrelated OpenClaw/WeChat report was traced to notification/task isolation weaknesses rather than GitOps baseline corruption.
 Actions completed:
 1. The legacy `AIOS每日自治报告` cron was disabled and retained for forensic review.
 2. The expired `2026-07-06-reminder` cron was archived and deleted after confirming no pending dependency.
 3. Pending notification queue depth was verified as zero.
 4. Unknown/unscoped queued message count was verified as zero.
 5. A scoped V1R4C6 WeChat test notification was sent successfully.
 The test notification included:
 ```text
 task_id = V1R4C6-integration
 release_id = V1R4C6
 branch = release/v1r4c6-admin-search-agent-integration-20260711
 baseline = V1R4C5
 correlation_id = 70308f5e243ab81d
 ```
 No recipient identifier, token, cookie, or unrelated message content is stored in this release record.
 ## Resume Decision
 V1R4C6 may resume:
 ```text
 merge
 component tests
 security/isolation tests
 deployment
 production verification
 release evidence updates
 ```
 V1R4C6 must still stop before:
 ```text
 Git tag creation
 baseline promotion
 memory search activation
 ```
 ## Required Execution Context
 Before every OpenClaw execution or WeChat notification, assert:
 ```text
 task_id = V1R4C6-integration
 release_id = V1R4C6
 branch = release/v1r4c6-admin-search-agent-integration-20260711
 baseline = V1R4C5
 memory_search_enabled = false
 ```
 Any mismatch re-enters safe mode and blocks execution/notification.
 ## Hard Constraints
 ```text
 no force-push
 no Gitea DB mutation
 no V1R4C5 tag movement
 no global memory/vector index
 no cross-user/project/session notification or task data
 ```
 ## Owner Gate
 After Gate A, Gate B, Gate C, deployment verification, runtime acceptance, and Gitea sentinel checks pass, OpenClaw must request explicit owner approval through the scoped WeChat channel for both:
 1. Create V1R4C6 Git tag.
 2. Promote active baseline from V1R4C5 to V1R4C6.
 Ambiguous acknowledgements do not count as approval.
@@ -1,136 +0,0 @@
 # V1R4C6 Integration Merge Matrix
 Date: 2026-07-11
 Status: integration-ready
 Base baseline: V1R4C5
 Base baseline git: `a32c76a3c1baa49562e53fb8adce10a5ed0e6704`
 ## Release Objective
 Deliver three capability groups in one controlled release train:
 1. Admin Workbench and Portal AIOS read-only rollout.
 2. SR-03 Search Gateway real provider binding.
 3. Agent Platform Integration.
 This is a bundled release train, not an uncontrolled bulk merge. Each group has an independent acceptance gate and rollback switch.
 ## Group A — Admin Workbench and Portal AIOS
 | Branch | Capability | Decision |
 |---|---|---|
 | `feature/v1r4c4-sr04-admin-workbench-visibility-20260711` | Admin workbench task progress and process visibility | integrate |
 | `feature/v1r4c4-sr04-portal-aios-visibility-20260710` | Portal AIOS route/menu visibility | integrate |
 | `patch/user-portal-me-route-diagnostics-20260709` | `/portal/me/aios/manifest`, dashboard and health route contract | integrate after route review |
 | `patch/user-portal-rbac-ops-dashboard` | Admin/group-admin RBAC dashboard | integrate after RBAC review |
 | `patch/agent-observability-webui` | Agent observability UI | integrate read-only |
 | `patch/web-ui-complete-gitops-dashboard-20260706` | GitOps dashboard UI | integrate only if not superseded by the newer workbench implementation |
 ### Required user-visible result
 Administrators can see:
 - task progress;
 - task/flow/process diagram entry;
 - system health;
 - authorized agent observability;
 - explicit empty/error states.
 ### Security constraints
 - Read-only first.
 - RBAC is enforced server-side and client-side.
 - No cross-user, cross-project, cross-session, or cross-tenant data exposure.
 - Owner-only modules remain hidden from non-owner administrators.
 ## Group B — SR-03 Search Gateway Real Provider Binding
 | Branch | Capability | Decision |
 |---|---|---|
 | `feature/v1r4c4-sr02-search-gateway-resilience-20260710` | Gateway resilience and safe fallback | integrate |
 | `patch/search-gateway-governed-general-capability` | Governed provider binding and policy surface | integrate after scope review |
 | `patch/gitops-change-archive-audit` | Change/archive audit evidence | integrate or retain as supporting audit module |
 ### Required result
 - Real provider binding works.
 - Query scope is enforced before ranking.
 - Accepted records include provenance.
 - Provider failure falls back to `text-only`, `no-memory`, or `scoped-only`.
 - Global memory/vector indexes are not used.
 ### Security constraints
 ```text
 AIOS_MEMORY_ENABLED=false
 ```
 Search Gateway provider binding must not implicitly enable memory search.
 ## Group C — Agent Platform Integration
 | Branch | Capability | Decision |
 |---|---|---|
 | `patch/agent-event-bus-unified` | Unified scoped agent event bus | integrate |
 | `patch/agent-gpt-message-plane` | Scoped Agent/GPT message plane | integrate |
 | `patch/autonomous-controller-dispatch-20260709` | Autonomous controller dispatch | integrate behind safe-mode gate |
 | `patch/self-evolution-agent-orchestration-policy` | Agent orchestration and governance policy | integrate |
 | `patch/aios-autonomous-orchestrator-architecture-review-v1` | Architecture record | integrate as documentation or mark superseded after review |
 ### Required result
 - Agent event and message envelopes include user/project/session scope.
 - Controller dispatch stops when autonomous safe mode is active.
 - Consequential actions remain owner-gated.
 - Observability is available before autonomous expansion.
 ## Explicitly Excluded
 Do not merge or activate in V1R4C6:
 - memory search activation;
 - global vector index;
 - cross-session/global recall;
 - unscoped search or message delivery;
 - old test/smoke/auth-only branches;
 - old V1R2/V1R3 promotion requests;
 - force-push reconciliation;
 - Gitea database mutation.
 ## Merge and Deployment Order
 ```text
 1. Verify V1R4C5 baseline and bare-repo configuration.
 2. Integrate Group A read-only workbench/portal surface.
 3. Run Group A RBAC and visibility acceptance.
 4. Integrate Group B provider binding under isolation gate.
 5. Run Group B scope/provenance/fallback acceptance.
 6. Integrate Group C event/message/dispatch capabilities.
 7. Run Group C scope/safe-mode/owner-gate acceptance.
 8. Run combined runtime acceptance.
 9. Prepare V1R4C6 tag and promotion request.
 10. Stop for owner confirmation before tag/promotion.
 ```
 ## Combined Acceptance
 Required:
 ```text
 getRuntimeStatus -> active
 getBaseline -> V1R4C5 before promotion
 createBranch -> committed and synced
 createPatch -> committed and synced
 Gitea branch sentinel -> healthy
 memory search -> disabled
 ```
 ## Rollback Switches
 - Group A: disable new workbench routes/menu entries.
 - Group B: disable real provider binding and use safe non-memory fallback.
 - Group C: disable expanded dispatch and keep event/message plane observational.
 - Memory search remains disabled in all rollback modes.
@@ -1,51 +0,0 @@
 # V1R4C6 Release Acceptance Checklist
 Date: 2026-07-11
 ## Gate A — Admin Workbench / Portal AIOS
 | Check                          | Status  | Notes                          |
 |--------------------------------|---------|--------------------------------|
 | Admin workbench RBAC read-only | NOT RUN |                                |
 | Portal AIOS manifest route     | NOT RUN |                                |
 | Portal AIOS dashboard route    | NOT RUN |                                |
 | Flow/process diagram renders   | NOT RUN |                                |
 | Portal AIOS health route       | NOT RUN |                                |
 ## Gate B — SR-03 Search Gateway
 | Check                              | Status  | Notes                          |
 |------------------------------------|---------|--------------------------------|
 | Provider binding correctness       | NOT RUN |                                |
 | Scope-before-ranking validation    | NOT RUN |                                |
 | Provenance correctness             | NOT RUN |                                |
 | Fallback to disabled provider      | NOT RUN |                                |
 ## Gate C — Agent Platform
 | Check                              | Status  | Notes                          |
 |------------------------------------|---------|--------------------------------|
 | Event scope isolation               | NOT RUN |                                |
 | Message plane scope isolation       | NOT RUN |                                |
 | Dispatch safe-mode respects control | NOT RUN |                                |
 | Owner gate for consequential ops    | NOT RUN |                                |
 ## Final Deployment Gates
 | Check                     | Status    | Notes                          |
 |---------------------------|-----------|--------------------------------|
 | Tag V1R4C6                | NO        | Requires owner approval        |
 | Baseline promotion        | NO        | Requires owner approval        |
 | Memory search enabled     | NO        | Requires future owner approval |
 ## Active Constraints
 - AIOS_MEMORY_ENABLED: false
 - Force-push: forbidden
 - Gitea DB mutation: owner approval required
 - Memory search: disabled (future owner approval required for enable)
 ## Artifact
 - **PR #6 merge commit**: `ae49bf0fa35b0f599a96a4d12644576447f39cec`
 - **Source baseline**: V1R4C5 (commit a32c76a3c1baa49562e53fb8adce10a5ed0e6704)
@@ -1,92 +0,0 @@
 # V1R4C6 Release Gate Status
 Date: 2026-07-11
 Status: integration pending — release-candidate control artifacts merged, engineering gates not yet executed
 ## Completed
 - V1R4C5 baseline verified as active base.
 - GitOps API and public API return:
  - current_baseline: V1R4C5
  - previous_baseline: V1R4C3
  - git: a32c76a3c1baa49562e53fb8adce10a5ed0e6704
 - GitOps bare repository path corrected to active `aios.git`.
 - V1R4C6 merge matrix created.
 - V1R4C6 acceptance checklist created.
 - V1R4C6 acceptance evidence created.
 - PR #6: control artifacts merge — PASS
 ## Result Summary
 | Item                    | Status    |
 |-------------------------|-----------|
 | control artifacts merge | PASS       |
 | Gate A                  | NOT RUN   |
 | Gate B                  | NOT RUN   |
 | Gate C                  | NOT RUN   |
 | tests                   | NOT RUN   |
 | artifact                | NOT CREATED |
 | deployment              | NOT RUN   |
 | production verification | NOT RUN   |
 | tag created             | NO        |
 | baseline promoted       | NO        |
 | memory search enabled   | NO        |
 | **overall**             | **BLOCKED / INTEGRATION PENDING** |
 ## Verified Artifact
 - **PR #6 merge commit**: `ae49bf0fa35b0f599a96a4d12644576447f39cec`
 - **Current baseline**: V1R4C5
 - **V1R4C6 tag**: absent
 - **Memory search**: disabled
 - **Force-push**: disabled
 ## Remaining Gates (not yet executed)
 ### Gate A — Admin Workbench / Portal AIOS
 - merge candidate review;
 - RBAC acceptance;
 - task progress visibility;
 - flow diagram visibility;
 - health/observability validation.
 ### Gate B — SR-03 Search Gateway
 - provider binding validation;
 - scope-before-ranking validation;
 - provenance validation;
 - fallback validation.
 ### Gate C — Agent Platform
 - event scope validation;
 - message plane scope validation;
 - dispatch safe-mode validation;
 - owner gate validation.
 ## Release Stop Point
 After Gates A/B/C pass:
 ```text
 prepare V1R4C6 tag
        ↓
 owner confirmation required
        ↓
 tag creation
        ↓
 baseline promotion
 ```
 ## Hard Constraints
 Remain:
 ```text
 AIOS_MEMORY_ENABLED=false
 memory search disabled
 force-push forbidden
 gitea DB mutation requires owner approval
 ```
@@ -1,67 +0,0 @@
 # Weixin Quoted Text Fail-Closed Implementation Plan
 > **For agentic workers:** REQUIRED SUB-SKILL: Use superpowers:subagent-driven-development to implement this plan task-by-task. Steps use checkbox (`- [ ]`) syntax for tracking.
 **Goal:** Resolve explicit quoted text safely and fail closed without a model call when Weixin omits the quoted body.
 **Architecture:** Add pure JS/TS reference-text helpers and one process hook guard to the existing patcher. Keep the patch surface limited to generated `process-message.ts/js`; do not introduce a persistent cache or modify send/monitor/API modules.
 **Tech Stack:** Python patch generator, TypeScript/JavaScript, Node test harness, pytest.
 ## Global Constraints
 - Sub-agents use `gpt-5.3-codex-spark`.
 - No force push, tag, baseline promotion, credential persistence, recent-message fallback, or approximate text matching.
 - Existing confirmed image and ordinary text behavior must remain unchanged.
 ---
 ### Task 1: Add Quoted Text Resolver With TDD
 **Files:**
 - Modify: `tests/test_weixin_image_followup_intent.py`
 - Modify: `fixes/patch_openclaw_weixin_hermes.py`
 **Interfaces:**
 - Produces: `extractHermesReferencedText(items)` and `resolveHermesQuotedTextContext(items, currentText)` in TS and JS helper templates.
 - [ ] Add a Node-backed failing test with `ref_msg.message_item.type=1` and explicit `text_item.text="被引用的正文"`; assert `reason="quoted_text_resolved"`, `handled=false`, and a prompt containing only delimited quoted/current text.
 - [ ] Add a Node-backed failing test with `type=0`, quoted ID, no text, and no confirmed image; assert `reason="quoted_text_unavailable"`, `handled=true`, and exact deterministic `replyText`.
 - [ ] Run `pytest -q tests/test_weixin_image_followup_intent.py -k 'quoted_text'` and record expected RED because the resolver is absent.
 - [ ] Implement the smallest recursive explicit-text extractor. Accept only known referenced text fields; cap each extracted string at 4000 characters and reject blank values.
 - [ ] Implement the resolver in both TS and JS templates. Do not read session history, image latest cache, timestamps, or metadata values.
 - [ ] Re-run the focused tests and record GREEN.
 - [ ] Commit with `fix(weixin): fail closed on unavailable quoted text`.
 ### Task 2: Guard The Real Process Hook
 **Files:**
 - Modify: `tests/test_weixin_image_followup_intent.py`
 - Modify: `fixes/patch_openclaw_weixin_hermes.py`
 **Interfaces:**
 - Consumes: `resolveHermesQuotedTextContext(items, currentText)`.
 - Produces: process hook behavior that replies deterministically or replaces only the ordinary text prompt.
 - [ ] Add a failing JS hook test that stubs the ordinary model dispatch and reply function. For type 0 without text, assert reply called once and model called zero times.
 - [ ] Add a failing JS hook test for explicit quoted text. Assert model called once with delimited context and no message metadata.
 - [ ] Add regression assertions for plain text, current image, and exact quoted image.
 - [ ] Run focused tests and record RED for the missing guard.
 - [ ] Insert the resolver after reference/image evidence extraction and before ordinary model dispatch. Confirmed image references must bypass the text fail-closed branch.
 - [ ] Emit structured reason-only diagnostics without message bodies.
 - [ ] Run focused tests and record GREEN.
 - [ ] Commit with `fix(weixin): guard quoted text model dispatch`.
 ### Task 3: Idempotence, Review, And Release
 **Files:**
 - Modify: `docs/incidents/WEIXIN_TEXT_IMAGE_CACHE_ISOLATION_20260711.md`
 - [ ] Run `pytest -q tests/test_weixin_image_followup_intent.py` and require all tests pass.
 - [ ] Run `python3 -m py_compile fixes/patch_openclaw_weixin_hermes.py tests/test_weixin_image_followup_intent.py`.
 - [ ] Run `git diff --check`.
 - [ ] Apply `--process-only` twice to a scratch production-plugin copy and require identical second-run hashes; require slash/channel files unchanged.
 - [ ] Obtain independent spec and code review.
 - [ ] Update the incident document with RED/GREEN evidence, main revision, production hashes, rollback directory, and owner live acceptance state.
 - [ ] Merge by non-force fast-forward from a clean main-based branch, deploy only the two generated process-message files, restart only `openclaw-gateway.service`, and verify service, port 18789, four accounts, zero post-start errors, and unchanged image cache.
 - [ ] Owner live acceptance: quote a type 0 text message and verify deterministic no-body response; quote a payload that includes explicit text and verify grounded explanation.
@@ -1,45 +0,0 @@
 # Weixin Quoted Text Fail-Closed Design
 ## Goal
 Prevent quoted-text questions from being answered with invented message metadata when Weixin does not provide the quoted body. Preserve normal text and confirmed image behavior.
 ## Evidence
 - A quoted text event can arrive as `ref_msg.message_item.type=0` with only `msg_id` and timestamps.
 - Session transcripts retain text but not the original Weixin numeric message ID.
 - Outbound `clientId` values are not the same format as quoted numeric `msg_id` values, and the send API does not expose a server message ID.
 - Therefore recent-message, timestamp, session, and outbound-client-ID matching are not exact evidence and must not be used.
 ## Decision
 Use an evidence-only resolver in `process-message`:
 1. If reference payload contains an explicit text item, extract that text and build a bounded quoted-text context for the ordinary text model path.
 2. If reference context exists but has no explicit text and is not a confirmed image, return a deterministic response: `无法读取被引用文字，请复制原文后再提问。`
 3. The unresolved branch is handled before ordinary model dispatch and before Hermes image dispatch.
 4. Do not infer quoted text from metadata, recent messages, session order, timestamps, or approximate IDs.
 No persistent text cache is introduced in this change. Outbound ID aliasing remains a separate observation task because an exact mapping has not been proved.
 ## Interfaces
 - `extractHermesReferencedText(items): string` returns only explicit referenced text, otherwise an empty string.
 - `resolveHermesQuotedTextContext(items, currentText)` returns one of:
  - `{ referenced: false, handled: false, prompt: currentText, reason: "none" }`
  - `{ referenced: true, handled: false, prompt, reason: "quoted_text_resolved" }`
  - `{ referenced: true, handled: true, replyText: "无法读取被引用文字，请复制原文后再提问。", reason: "quoted_text_unavailable" }`
 - Confirmed referenced images bypass this text resolver and keep the existing exact image path.
 ## Privacy And Logging
 - Do not log quoted or current text.
 - Log only `reason`, account ID, a one-way bounded hash of quoted ID when available, and boolean evidence flags.
 - Do not persist tokens, credentials, full payloads, or message text.
 ## Acceptance
 - Explicit quoted text is supplied to the ordinary text model with a clear delimiter.
 - Type 0 reference without text returns the deterministic Chinese response and does not call the model.
 - Plain text, current image, and exact quoted image behavior remain unchanged.
 - Generated TS and JS remain byte-idempotent under `--process-only`.
@@ -1,152 +0,0 @@
 apiVersion: aios.gitops/v1
 kind: ExecutionOrder
 metadata:
  name: openclaw-v1r4c6-merge-tests-deploy
  date: 2026-07-11
  owner_approval:
    branch: request/openclaw-v1r4c6-merge-deploy-20260711
    file: approvals/openclaw-v1r4c6-merge-deploy-owner-approval.yaml
    commit: 4653f92
 spec:
  objective: >-
    Integrate the approved V1R4C6 Admin Workbench, Search Gateway, and Agent Platform
    changes into a clean candidate based on the latest main, execute all mandatory gates,
    merge to main only after success, then deploy and verify production.
  confirmation_policy: no_additional_confirmation_required
  source:
    primary_branch: release/v1r4c6-admin-search-agent-integration-20260711
    supporting_domains:
      - admin-workbench-portal-aios
      - search-gateway-governance
      - agent-platform-integration
  target:
    integration_branch: integration/v1r4c6-merge-deploy-20260711-01
    final_branch: main
    release_version: V1R4C6
  execution:
    - id: preflight
      actions:
        - fetch_all_refs_and_tags
        - verify_workspace_clean
        - reset_integration_candidate_to_latest_main
        - verify_baseline_is_V1R4C5
        - verify_memory_search_enabled_is_false
        - verify_V1R4C5_tag_unchanged
        - record_main_head_and_source_head
      stop_conditions:
        - dirty_workspace
        - missing_source_branch
        - baseline_not_V1R4C5
        - protected_setting_changed
    - id: semantic-diff-review
      actions:
        - compute_three_dot_diff_from_main_to_primary_branch
        - classify_changes_by_admin_search_agent_domains
        - detect_duplicate_control_artifacts_already_merged_by_PR_6
        - preserve_PR_7_NOT_RUN_status_until_tests_execute
        - reject_reversion_of_V1R4C5_baseline_metadata
        - reject_enablement_of_memory_search
      output:
        - execution/evidence/v1r4c6-semantic-diff-report.md
    - id: merge-candidate
      actions:
        - merge_primary_branch_into_integration_candidate_without_fast_forward
        - resolve_text_conflicts_by_preserving_latest_main_control_metadata
        - preserve_source_business_functionality_when_not_conflicting_with_safety_controls
        - commit_resolved_candidate
      forbidden:
        - force_push
        - direct_commit_to_main
        - bypass_branch_protection
        - bypass_owner_gate
        - weaken_rbac_or_scope_isolation
        - mark_unexecuted_tests_as_passed
    - id: gate-a-admin
      required_checks:
        - admin_workbench_routes_and_assets_load
        - portal_aios_manifest_dashboard_health_endpoints
        - server_side_and_frontend_rbac
        - empty_error_and_loading_states
        - tenant_user_project_session_isolation
    - id: gate-b-search
      required_checks:
        - real_provider_binding_or_explicit_safe_fallback
        - timeout_limited_retry_and_safe_fallback
        - scope_filter_before_ranking
        - provenance_in_results
        - provider_and_fallback_audit
        - unauthorized_results_never_enter_ranking_or_summary
        - memory_search_remains_disabled
    - id: gate-c-agent
      required_checks:
        - unified_agent_event_bus
        - scoped_agent_gpt_message_plane
        - autonomous_controller_dispatch
        - safe_mode_stops_dispatch_and_change
        - high_risk_actions_remain_owner_gated
        - duplicate_dispatch_is_idempotent_or_rejected
        - queue_failure_retry_observability
    - id: runtime-and-regression
      actions:
        - run_getRuntimeStatus
        - run_getBaseline
        - run_createBranch_smoke_test
        - run_createPatch_smoke_test
        - run_full_regression_suite
        - run_deployment_preflight_health_checks
      expected:
        reads: "200 OK with service, status, workspace_branch"
        writes: "200 OK with branch, commit, rev, sync_status"
    - id: promote-main
      conditions:
        - gate_a_passed
        - gate_b_passed
        - gate_c_passed
        - runtime_acceptance_passed
        - regression_passed
        - deployment_preflight_passed
      actions:
        - open_or_update_merge_request_from_integration_candidate_to_main
        - merge_without_rewriting_history
        - record_new_main_head
        - update_versions_V1R4C6_json_to_release_ready_or_active_baseline_as_policy_requires
        - update_baseline_current_json_only_after_successful_promotion
        - create_V1R4C6_tag_according_to_repository_tag_semantics
    - id: deploy-and-verify
      actions:
        - trigger_direct_deployment_from_new_main
        - verify_service_health
        - verify_admin_search_agent_smoke_paths
        - verify_runtime_status_and_baseline_after_deploy
        - verify_rollback_artifact_and_previous_baseline_are_available
        - publish_production_verification_report
  failure_policy:
    mode: fail_closed
    on_any_failure:
      - stop_pipeline
      - do_not_update_main
      - do_not_create_or_move_release_tag
      - do_not_deploy
      - preserve_integration_branch_for_forensics
      - commit_failure_evidence_and_exact_blocker
  reporting:
    required_files:
      - execution/evidence/v1r4c6-semantic-diff-report.md
      - execution/evidence/gate-a-admin-result.md
      - execution/evidence/gate-b-search-result.md
      - execution/evidence/gate-c-agent-result.md
      - execution/evidence/runtime-acceptance-result.md
      - execution/evidence/regression-result.md
      - execution/evidence/deployment-result.md
      - execution/evidence/production-verification-result.md
    final_summary_fields:
      - source_head
      - integration_head
      - main_head_before
      - main_head_after
      - merge_commit
      - release_tag
      - deployment_id
      - gate_results
      - production_health
      - rollback_ready
@@ -1,112 +0,0 @@
 from __future__ import annotations
 import argparse
 import json
 from pathlib import Path
 TS_HELPER = '''// AIOS Weixin lossless getUpdates JSON begin
 const WEIXIN_EXACT_ID_FIELDS = new Set(["message_id", "msg_id", "root_id", "parent_id", "server_msg_id", "new_msg_id"]);
 function parseWeixinGetUpdatesJson(rawText: string): GetUpdatesResp {
  return JSON.parse(rawText, (key: string, value: unknown, context?: { source?: string }) => {
    if (!WEIXIN_EXACT_ID_FIELDS.has(key) || value == null || typeof value === "string") return value;
    if (typeof value !== "number") throw new Error(`invalid ${key} type in getUpdates response`);
    const source = context?.source;
    if (!source || !/^(?:0|[1-9]\\d*)$/.test(source)) throw new Error(`unsafe numeric ${key} in getUpdates response`);
    return source;
  }) as GetUpdatesResp;
 }
 // AIOS Weixin lossless getUpdates JSON end'''
 JS_HELPER = '''// AIOS Weixin lossless getUpdates JSON begin
 const WEIXIN_EXACT_ID_FIELDS = new Set(["message_id", "msg_id", "root_id", "parent_id", "server_msg_id", "new_msg_id"]);
 function parseWeixinGetUpdatesJson(rawText) {
    return JSON.parse(rawText, (key, value, context) => {
        if (!WEIXIN_EXACT_ID_FIELDS.has(key) || value == null || typeof value === "string")
            return value;
        if (typeof value !== "number")
            throw new Error(`invalid ${key} type in getUpdates response`);
        const source = context?.source;
        if (!source || !/^(?:0|[1-9]\\d*)$/.test(source))
            throw new Error(`unsafe numeric ${key} in getUpdates response`);
        return source;
    });
 }
 // AIOS Weixin lossless getUpdates JSON end'''
 def insert_once(text: str, marker: str, addition: str) -> str:
    if addition in text:
        return text
    if marker not in text:
        raise RuntimeError(f'marker not found: {marker}')
    return text.replace(marker, marker + '\n' + addition, 1)
 def patch_api_ts(text: str) -> str:
    text = insert_once(text, 'import { redactBody, redactUrl } from "../util/redact.js";', '\n' + TS_HELPER)
    old = 'const resp: GetUpdatesResp = JSON.parse(rawText);'
    new = 'const resp: GetUpdatesResp = parseWeixinGetUpdatesJson(rawText);'
    if new not in text:
        if old not in text:
            raise RuntimeError('TypeScript getUpdates parser marker not found')
        text = text.replace(old, new, 1)
    return text
 def patch_api_js(text: str) -> str:
    text = insert_once(text, 'import { redactBody, redactUrl } from "../util/redact.js";', '\n' + JS_HELPER)
    old = 'const resp = JSON.parse(rawText);'
    new = 'const resp = parseWeixinGetUpdatesJson(rawText);'
    if new not in text:
        if old not in text:
            raise RuntimeError('JavaScript getUpdates parser marker not found')
        text = text.replace(old, new, 1)
    return text
 def patch_types_ts(text: str) -> str:
    new = 'message_id?: string | number;'
    if new in text:
        return text
    old = 'message_id?: number;'
    if old not in text:
        raise RuntimeError('message_id type marker not found')
    return text.replace(old, new, 1)
 def write_if_changed(path: Path, content: str) -> bool:
    original = path.read_text(encoding='utf-8')
    if original == content:
        return False
    path.write_text(content, encoding='utf-8')
    return True
 def apply(plugin_root: Path) -> dict[str, bool]:
    paths = {
        'src/api/api.ts': patch_api_ts,
        'dist/src/api/api.js': patch_api_js,
        'src/api/types.ts': patch_types_ts,
    }
    changed: dict[str, bool] = {}
    for relative, transform in paths.items():
        path = plugin_root / relative
        if not path.exists():
            raise RuntimeError(f'missing plugin file: {path}')
        original = path.read_text(encoding='utf-8')
        changed[relative] = write_if_changed(path, transform(original))
    return changed
 def main() -> int:
    parser = argparse.ArgumentParser()
    parser.add_argument('plugin_root', type=Path)
    args = parser.parse_args()
    changed = apply(args.plugin_root.resolve())
    print(json.dumps({'plugin_root': str(args.plugin_root.resolve()), 'changed': changed}, ensure_ascii=False))
    return 0
 if __name__ == '__main__':
    raise SystemExit(main())
@@ -1,156 +0,0 @@
 from __future__ import annotations
 import argparse
 import json
 from pathlib import Path
 def replace_required(text: str, old: str, new: str, label: str) -> str:
    if new in text:
        return text
    if old not in text:
        raise RuntimeError(f'{label} marker not found')
    return text.replace(old, new, 1)
 def patch_send_ts(text: str) -> str:
    text = replace_required(
        text,
        '}): Promise<{ messageId: string }> {\n  const { to, text, opts } = params;\n',
        '}): Promise<{ messageId: string; createTimeMs: number }> {\n  const { to, text, opts } = params;\n',
        'TS send return type',
    )
    text = replace_required(
        text,
        '  const clientId = generateClientId();\n  const req = buildSendMessageReq({',
        '  const clientId = generateClientId();\n  const createTimeMs = Date.now();\n  const req = buildSendMessageReq({',
        'TS create time',
    )
    text = replace_required(
        text,
        '  });\n  try {\n    await sendMessageApi({',
        '  });\n  if (!req.msg?.item_list?.[0]) throw new Error("sendMessageWeixin: text item missing");\n  req.msg.item_list[0].create_time_ms = createTimeMs;\n  try {\n    await sendMessageApi({',
        'TS request timestamp',
    )
    text = replace_required(
        text,
        '  return { messageId: clientId };\n}\n\n/** Send a single structured MessageItem downstream. */',
        '  return { messageId: clientId, createTimeMs };\n}\n\n/** Send a single structured MessageItem downstream. */',
        'TS send result',
    )
    return text
 def patch_send_js(text: str) -> str:
    text = replace_required(
        text,
        '    const clientId = generateClientId();\n    const req = buildSendMessageReq({',
        '    const clientId = generateClientId();\n    const createTimeMs = Date.now();\n    const req = buildSendMessageReq({',
        'JS create time',
    )
    text = replace_required(
        text,
        '    });\n    try {\n        await sendMessageApi({',
        '    });\n    if (!req.msg?.item_list?.[0])\n        throw new Error("sendMessageWeixin: text item missing");\n    req.msg.item_list[0].create_time_ms = createTimeMs;\n    try {\n        await sendMessageApi({',
        'JS request timestamp',
    )
    text = replace_required(
        text,
        '    return { messageId: clientId };\n}\n/** Send a single structured MessageItem downstream. */',
        '    return { messageId: clientId, createTimeMs };\n}\n/** Send a single structured MessageItem downstream. */',
        'JS send result',
    )
    return text
 def patch_process_ts(text: str) -> str:
    old = '''await sendMessageWeixin({ to: ctx.To, text, opts: {
                baseUrl: deps.baseUrl,
                token: deps.token,
                contextToken,
                runId,
              }});'''
    new = '''const sentText = await sendMessageWeixin({ to: ctx.To, text, opts: {
                baseUrl: deps.baseUrl,
                token: deps.token,
                contextToken,
                runId,
              }});
              rememberHermesOutboundTextMessage(deps.accountId, ctx.To, text, sentText.createTimeMs);'''
    text = replace_required(text, old, new, 'TS fallback text send')
    old = '''await sendMessageWeixin({ to: ctx.To, text, opts: {
              baseUrl: deps.baseUrl,
              token: deps.token,
              contextToken,
              runId,
            }});'''
    new = '''const sentText = await sendMessageWeixin({ to: ctx.To, text, opts: {
              baseUrl: deps.baseUrl,
              token: deps.token,
              contextToken,
              runId,
            }});
            rememberHermesOutboundTextMessage(deps.accountId, ctx.To, text, sentText.createTimeMs);'''
    return replace_required(text, old, new, 'TS normal text send')
 def patch_process_js(text: str) -> str:
    old = '''await sendMessageWeixin({ to: ctx.To, text, opts: {
                                baseUrl: deps.baseUrl,
                                token: deps.token,
                                contextToken,
                                runId,
                            } });'''
    new = '''const sentText = await sendMessageWeixin({ to: ctx.To, text, opts: {
                                baseUrl: deps.baseUrl,
                                token: deps.token,
                                contextToken,
                                runId,
                            } });
                        rememberHermesOutboundTextMessage(deps.accountId, ctx.To, text, sentText.createTimeMs);'''
    text = replace_required(text, old, new, 'JS fallback text send')
    old = '''await sendMessageWeixin({ to: ctx.To, text, opts: {
                            baseUrl: deps.baseUrl,
                            token: deps.token,
                            contextToken,
                            runId,
                        } });'''
    new = '''const sentText = await sendMessageWeixin({ to: ctx.To, text, opts: {
                            baseUrl: deps.baseUrl,
                            token: deps.token,
                            contextToken,
                            runId,
                        } });
                    rememberHermesOutboundTextMessage(deps.accountId, ctx.To, text, sentText.createTimeMs);'''
    return replace_required(text, old, new, 'JS normal text send')
 def apply(root: Path) -> dict[str, bool]:
    transforms = {
        'src/messaging/send.ts': patch_send_ts,
        'dist/src/messaging/send.js': patch_send_js,
        'src/messaging/process-message.ts': patch_process_ts,
        'dist/src/messaging/process-message.js': patch_process_js,
    }
    changed = {}
    for relative, transform in transforms.items():
        path = root / relative
        original = path.read_text(encoding='utf-8')
        updated = transform(original)
        changed[relative] = updated != original
        if updated != original:
            path.write_text(updated, encoding='utf-8')
    return changed
 def main() -> int:
    parser = argparse.ArgumentParser()
    parser.add_argument('plugin_root', type=Path)
    args = parser.parse_args()
    root = args.plugin_root.resolve()
    print(json.dumps({'plugin_root': str(root), 'changed': apply(root)}, ensure_ascii=False))
    return 0
 if __name__ == '__main__':
    raise SystemExit(main())
@@ -1,429 +0,0 @@
 # CMD-20260711-v1r4c6-integrate-deploy-verify
 ## Goal
 Perform the actual controlled integration, deployment, and verification for V1R4C6.
 V1R4C6 includes:
 1. Admin Workbench / Portal AIOS read-only rollout.
 2. SR-03 Search Gateway real provider binding.
 3. Agent Platform Integration.
 Stop before V1R4C6 Git tag creation and baseline promotion. Those operations require owner confirmation.
 ## Hard Constraints
 ```text
 AIOS_MEMORY_ENABLED=false
 memory search remains disabled
 no global vector index
 no force-push
 no V1R4C5 tag movement
 no Gitea DB mutation
 no secret plaintext in Git, logs, or reports
 ```
 If any mandatory security or reliability check fails, stop deployment and report the blocker. Do not bypass the gate.
 ## Verified Base
 Required before starting:
 ```text
 current_baseline = V1R4C5
 previous_baseline = V1R4C3
 baseline_git = a32c76a3c1baa49562e53fb8adce10a5ed0e6704
 AIOS_GITOPS_BARE_REPO = /data/aios-gitops/runtime/gitea-bin/data/repositories/aios/aios.git
 runtime = active
 ```
 ## Phase 0 — Preflight and Backup
 ```bash
 set -euo pipefail
 ROOT=/data/aios-gitops/workspace
 SOURCE_RELEASE=release/v1r4c6-admin-search-agent-integration-20260711
 EXEC_BRANCH=release/v1r4c6-integration-exec-20260711
 cd "$ROOT"
 git fetch origin --tags --prune
 git status --short
 test -f baseline/current.json
 python3 - <<'PY'
 import json
 from pathlib import Path
 p = Path('baseline/current.json')
 d = json.loads(p.read_text(encoding='utf-8'))
 assert d['current_baseline'] == 'V1R4C5', d
 assert d['previous_baseline'] == 'V1R4C3', d
 assert d['git'] == 'a32c76a3c1baa49562e53fb8adce10a5ed0e6704', d
 assert d.get('constraints', {}).get('memory_search_enabled') is False, d
 print('baseline preflight: PASS')
 PY
 test -d /data/aios-gitops/runtime/gitea-bin/data/repositories/aios/aios.git
 BACKUP=/data/aios-gitops/runtime/backups/v1r4c6-predeploy-$(date +%Y%m%d-%H%M%S)
 mkdir -p "$BACKUP"
 cp -a baseline/current.json "$BACKUP/current.json"
 cp -a versions/V1R4C5.json "$BACKUP/V1R4C5.json"
 if git show-ref --verify --quiet "refs/heads/$EXEC_BRANCH"; then
  git checkout "$EXEC_BRANCH"
  git pull --ff-only origin "$EXEC_BRANCH"
 else
  git checkout -b "$EXEC_BRANCH" "origin/$SOURCE_RELEASE"
  git push -u origin "$EXEC_BRANCH"
 fi
 ```
 Never use `git push --force` or `git tag -f`.
 ## Phase 1 — Merge Group A: Admin Workbench / Portal AIOS
 Merge in this order:
 ```bash
 git merge --no-ff origin/feature/v1r4c4-sr04-admin-workbench-visibility-20260711 \
  -m 'merge(v1r4c6): admin workbench visibility preparation'
 git merge --no-ff origin/feature/v1r4c4-sr04-portal-aios-visibility-20260710 \
  -m 'merge(v1r4c6): portal AIOS visibility preparation'
 git merge --no-ff origin/patch/user-portal-me-route-diagnostics-20260709 \
  -m 'merge(v1r4c6): portal AIOS routes and RBAC manifest'
 git merge --no-ff origin/patch/user-portal-rbac-ops-dashboard \
  -m 'merge(v1r4c6): portal RBAC operations dashboard'
 git merge --no-ff origin/patch/agent-observability-webui \
  -m 'merge(v1r4c6): agent observability web UI'
 ```
 Before merging `patch/web-ui-complete-gitops-dashboard-20260706`, compare overlap:
 ```bash
 git diff --stat HEAD..origin/patch/web-ui-complete-gitops-dashboard-20260706
 ```
 Merge only if it adds required functionality not already present. Otherwise record it as superseded.
 ### Conflict Policy
 Do not auto-resolve conflicts in:
 ```text
 auth/RBAC configuration
 user/session/tenant isolation
 search provider configuration
 safe-mode/controller dispatch
 systemd units
 baseline/current.json
 versions/*.json
 ```
 For those files, stop and report conflict paths.
 ### Group A Tests
 Run at minimum:
 ```bash
 python3 -m pytest -q \
  tests/test_portal_aios_frontend_contract.py \
  tests/test_portal_aios_module_registry.py \
  tests/test_portal_aios_navigation.py \
  tests/test_portal_integration_health.py \
  tests/test_portal_system_health_adapter.py
 ```
 Run any additional portal/RBAC/dashboard tests discovered under `tests/`.
 ## Phase 2 — Merge Group B: SR-03 Search Gateway
 ```bash
 git merge --no-ff origin/feature/v1r4c4-sr02-search-gateway-resilience-20260710 \
  -m 'merge(v1r4c6): search gateway resilience metadata'
 git merge --no-ff origin/patch/search-gateway-governed-general-capability \
  -m 'merge(v1r4c6): governed search gateway provider binding'
 git merge --no-ff origin/patch/gitops-change-archive-audit \
  -m 'merge(v1r4c6): search and GitOps audit evidence'
 ```
 ### Group B Security Review
 Verify before deployment:
 ```text
 scope filtering occurs before ranking
 provenance is attached to accepted results
 out-of-scope records never reach ranking or summarization
 fallback is text-only, no-memory, or scoped-only
 global memory/vector index is disabled
 memory search remains disabled
 provider credentials are secret references only
 ```
 Reject any code path that silently broadens search scope.
 ### Group B Tests
 Discover and run all search/provider/isolation tests:
 ```bash
 find tests -maxdepth 2 -type f \
  \( -iname '*search*' -o -iname '*provider*' -o -iname '*scope*' -o -iname '*isolation*' \) \
  -print
 ```
 Then run the discovered files with `python3 -m pytest -q`.
 Also run:
 ```bash
 bash ops/ci/run-user-data-isolation-acceptance.sh
 ```
 If that script is not present after merge, stop: the data-isolation gate is mandatory.
 ## Phase 3 — Merge Group C: Agent Platform
 ```bash
 git merge --no-ff origin/patch/agent-event-bus-unified \
  -m 'merge(v1r4c6): unified scoped agent event bus'
 git merge --no-ff origin/patch/agent-gpt-message-plane \
  -m 'merge(v1r4c6): scoped agent GPT message plane'
 git merge --no-ff origin/patch/autonomous-controller-dispatch-20260709 \
  -m 'merge(v1r4c6): autonomous controller dispatch with gates'
 git merge --no-ff origin/patch/self-evolution-agent-orchestration-policy \
  -m 'merge(v1r4c6): self-evolution orchestration policy'
 ```
 Review `patch/aios-autonomous-orchestrator-architecture-review-v1`:
 - merge if it contains a current architecture record not already present;
 - otherwise mark it superseded in the release report.
 ### Group C Security Review
 Verify:
 ```text
 event envelope includes tenant/group, user, project, session, agent, source and provenance
 message plane rejects cross-scope delivery
 denied messages never forward body/content
 controller dispatch checks autonomous safe mode before mutation
 consequential actions remain owner-gated
 duplicate dispatch is idempotent or rejected
 queue/retry behavior is bounded and observable
 ```
 ### Group C Tests
 Discover and run agent/event/message/dispatch/controller tests:
 ```bash
 find tests -maxdepth 2 -type f \
  \( -iname '*agent*' -o -iname '*event*' -o -iname '*message*' -o -iname '*dispatch*' -o -iname '*autonomous_controller*' \) \
  -print
 ```
 Run all discovered files with `python3 -m pytest -q`.
 Re-run the SR-02 watcher/controller suite and require all tests to pass.
 ## Phase 4 — Full Test and Static Validation
 ```bash
 python3 -m compileall -q active-runtime scripts tests
 python3 -m pytest -q
 ```
 If the full suite has unrelated historical failures, report them separately, but all V1R4C6-touched component tests must pass.
 Validate JSON/YAML configuration without printing secrets.
 Verify:
 ```bash
 git diff --check
 git status --short
 ```
 Commit integration result normally:
 ```bash
 git add -A
 git commit -m 'release: integrate V1R4C6 admin search and agent platform'
 git push origin "$EXEC_BRANCH"
 ```
 If no uncommitted changes remain because merge commits already captured everything, just push normally.
 ## Phase 5 — Deploy to Staging/Production Runtime
 Use the existing deployment mechanism for this repository. Do not invent a new deployment path.
 Before restart/reload, identify changed services and frontend bundles from the merged diff.
 At minimum:
 1. Deploy backend modules under `active-runtime/app/`.
 2. Deploy portal frontend/navigation/proxy changes.
 3. Deploy Search Gateway provider binding using secret references.
 4. Deploy Agent Platform event/message/dispatch changes.
 5. Preserve existing Gitea sentinel and user-data isolation systemd controls.
 6. Keep `AIOS_MEMORY_ENABLED=false`.
 Restart only affected services. Do not restart Gitea unless it independently fails health checks.
 After deployment, verify affected services are active and no restart loop exists.
 ## Phase 6 — Actual Deployment Verification
 ### Global
 ```bash
 curl -fsS http://127.0.0.1:18890/api/base | python3 -m json.tool
 curl -fsS https://cnnetmkt.top/api/base | python3 -m json.tool
 ```
 Before V1R4C6 promotion both must still report V1R4C5.
 Verify Gitea sentinel remains healthy and memory search remains disabled.
 ### Gate A — Portal/Admin Workbench
 Use an existing authenticated test mechanism or secret-backed admin session. Do not log cookies/tokens.
 Verify:
 ```text
 /portal/me/aios/manifest
 /portal/me/aios/dashboard
 /portal/me/aios/health
 ```
 For `user`, `group_admin/admin`, and `owner`, record only module names/status, not personal data.
 Required admin-visible functions:
 ```text
 task progress
 process/flow diagram entry
 system health
 agent observability
 ```
 Verify forbidden owner modules are hidden from non-owner admins.
 ### Gate B — Search Gateway
 Execute scoped test queries using non-sensitive fixtures.
 Verify:
 ```text
 real provider selected
 scope applied before ranking
 provenance returned
 out-of-scope fixture denied
 safe fallback works
 memory search remains disabled
 ```
 ### Gate C — Agent Platform
 Using synthetic non-sensitive events/messages:
 ```text
 same-scope delivery succeeds
 cross-user/project/session delivery is denied
 denied payload body is not forwarded
 safe-mode blocks controller mutation
 owner-gated action remains pending approval
 duplicate dispatch is idempotent/rejected
 ```
 ## Phase 7 — Release Evidence
 Update:
 ```text
 docs/releases/V1R4C6-acceptance-checklist.md
 docs/releases/V1R4C6-acceptance-evidence.md
 docs/releases/V1R4C6-release-gate-status.md
 versions/V1R4C6.json
 ```
 Set each gate to PASS/FAIL with concrete test commands and results.
 Required final state before requesting owner confirmation:
 ```text
 Gate A = PASS
 Gate B = PASS
 Gate C = PASS
 full touched-component tests = PASS
 runtime write acceptance = PASS
 Gitea sentinel = healthy
 memory search = disabled
 force-push used = no
 V1R4C6 tag = absent
 active baseline = V1R4C5
 ```
 Commit and push evidence normally to `$EXEC_BRANCH`.
 ## Stop Point — Owner Confirmation Required
 Do not create `V1R4C6` tag.
 Do not promote baseline to V1R4C6.
 Do not enable memory search.
 When all gates pass, notify the owner through the configured notification mechanism, preferably WeChat if configured. The notification must contain no secrets and must request confirmation for:
 ```text
 Create V1R4C6 Git tag
 Promote active baseline from V1R4C5 to V1R4C6
 ```
 Required report fields:
 ```text
 Status
 Integration branch and HEAD
 Merged branches
 Skipped/superseded branches
 Gate A result
 Gate B result
 Gate C result
 Test totals
 Deployment services/build versions
 Portal/admin verification result
 Search provider verification result
 Agent platform verification result
 Gitea sentinel status
 Memory search status
 Active baseline
 V1R4C6 tag status
 Rollback readiness
 Owner confirmation required
 ```
 If WeChat notification is not configured, write the exact notification message to:
 ```text
 /data/aios-gitops/runtime/state/V1R4C6-owner-confirmation-message.txt
 ```
 and report that path.
@@ -1,217 +0,0 @@
 # CMD-20260711-v1r4c6-run-to-owner-gate-wechat
 ## Execution Mode
 Execute immediately using the existing V1R4C6 integration command:
 ```text
 ops/openclaw/CMD-20260711-v1r4c6-integrate-deploy-verify.md
 ```
 Continue through actual merge, tests, deployment, and production verification.
 Stop only at the owner gate for:
 ```text
 V1R4C6 Git tag creation
 V1R4C6 baseline promotion
 ```
 Do not ask the owner to perform routine engineering or deployment work.
 ## Verified Starting State
 Required before starting:
 ```text
 runtime = active
 current_baseline = V1R4C5
 previous_baseline = V1R4C3
 baseline_git = a32c76a3c1baa49562e53fb8adce10a5ed0e6704
 AIOS_GITOPS_BARE_REPO = /data/aios-gitops/runtime/gitea-bin/data/repositories/aios/aios.git
 memory_search_enabled = false
 force_push_allowed = false
 ```
 If any starting condition differs, stop and send a WeChat blocker notification.
 ## Required Work
 Complete all phases in the referenced command:
 1. Create/update the normal non-force-pushed integration execution branch.
 2. Merge and validate Admin Workbench / Portal AIOS.
 3. Merge and validate SR-03 Search Gateway real provider binding.
 4. Merge and validate Agent Platform Integration.
 5. Run component and full touched-scope test suites.
 6. Deploy using the repository's existing deployment mechanism.
 7. Verify production routes, RBAC, scope isolation, provenance, safe mode, and owner gates.
 8. Update V1R4C6 release evidence and version metadata.
 9. Push all normal commits to origin.
 10. Keep V1R4C6 tag absent and baseline at V1R4C5 until owner approval.
 ## Hard Constraints
 ```text
 AIOS_MEMORY_ENABLED=false
 memory search remains disabled
 no global memory/vector index
 no force-push
 no tag rewrite
 no V1R4C5 tag movement
 no Gitea DB mutation
 no plaintext secrets in Git, logs, reports, or WeChat
 ```
 Never bypass a failed security, isolation, RBAC, safe-mode, or owner gate.
 ## WeChat Progress Notifications
 Use the configured OpenClaw WeChat notification channel.
 Send concise progress notifications at these milestones:
 ### Notification 1 — Integration Started
 ```text
 V1R4C6 集成已启动。
 范围：管理员工作台、SR-03 Search Gateway、Agent Platform。
 当前基线：V1R4C5。
 memory search 保持关闭。
 下一步：完成合并、测试和部署验证。
 ```
 ### Notification 2 — Gate A Complete
 Include only:
 ```text
 Gate A result: PASS/FAIL
 Admin task progress visibility: PASS/FAIL
 Flow diagram visibility: PASS/FAIL
 RBAC/isolation: PASS/FAIL
 Blockers, if any
 ```
 Do not include user records, tokens, cookies, or sensitive task content.
 ### Notification 3 — Gate B Complete
 Include only:
 ```text
 Gate B result: PASS/FAIL
 Real provider binding: PASS/FAIL
 Scope-before-ranking: PASS/FAIL
 Provenance: PASS/FAIL
 Safe fallback: PASS/FAIL
 Memory search: disabled
 Blockers, if any
 ```
 ### Notification 4 — Gate C Complete
 Include only:
 ```text
 Gate C result: PASS/FAIL
 Event scope: PASS/FAIL
 Message plane scope: PASS/FAIL
 Safe-mode dispatch block: PASS/FAIL
 Owner gate: PASS/FAIL
 Blockers, if any
 ```
 ## Final WeChat Owner Authorization Request
 Send only after all mandatory gates and deployment verification pass.
 Use this message:
 ```text
 V1R4C6 已完成实际合并、测试、部署和生产验证，现等待最终发布授权。
 版本范围：
 1. 管理员工作台 / Portal AIOS 上线
 2. SR-03 Search Gateway Real Provider Binding
 3. Agent Platform Integration
 验收结果：
 - Gate A（管理员工作台 / RBAC / 隔离）：PASS
 - Gate B（Provider / scope / provenance / fallback）：PASS
 - Gate C（Event Bus / Message Plane / Dispatch / Owner Gate）：PASS
 - Runtime 写链路：PASS
 - Gitea consistency sentinel：healthy
 - memory search：仍为 disabled
 - force-push：未使用
 - 当前 active baseline：V1R4C5
 - V1R4C6 tag：尚未创建
 请明确回复以下其中一项：
 A. 批准创建 V1R4C6 Git tag，并将 active baseline 从 V1R4C5 晋升至 V1R4C6。
 B. 暂不批准，保持 V1R4C5，并说明需要处理的问题。
 ```
 ## Authorization Rules
 Do not treat reactions, emojis, acknowledgements, or ambiguous replies as approval.
 Valid approval must explicitly authorize both:
 ```text
 create V1R4C6 Git tag
 promote baseline V1R4C5 -> V1R4C6
 ```
 If the owner approves only one operation, execute only the approved operation and stop before the other.
 Before executing a consequential operation, record authorization in:
 ```text
 docs/releases/V1R4C6-owner-approval.md
 ```
 Record only:
 ```text
 approved operation(s)
 approved_at
 approval channel = wechat
 message/conversation identifier if available
 owner identity reference without copying personal message content
 memory search remains disabled
 ```
 Do not store access tokens, cookies, phone numbers, or unrelated chat content.
 ## WeChat Failure Handling
 If the configured WeChat channel is unavailable:
 1. Do not claim that a message was sent.
 2. Stop before tag/promotion.
 3. Write the exact final authorization request to:
 ```text
 /data/aios-gitops/runtime/state/V1R4C6-owner-confirmation-message.txt
 ```
 4. Write notification failure details to:
 ```text
 /data/aios-gitops/runtime/state/V1R4C6-wechat-notification-failure.json
 ```
 5. Report the failure through the active GitOps task/status channel.
 ## Final Stop Condition
 Before explicit owner authorization, final state must remain:
 ```text
 current_baseline = V1R4C5
 V1R4C6 tag = absent
 memory search = disabled
 ```
 After owner authorization, follow the existing tag confirmation and baseline promotion controls. Do not bypass consequential-operation confirmation layers.
@@ -1,9 +1,9 @@
 {
  "type": "branch",
-  "branch": "deploy/v1r4c6-production-20260711",
+  "branch": "feature/v1r4c4-owner-maintenance-gate-v2-20260710",
  "base": "main",
  "created_by": "chatgpt-gitops-bot",
-  "created_at": "2026-07-11T13:10:51Z",
+  "created_at": "2026-07-10T10:32:35Z",
  "direct_main_push": false,
  "owner_approval_required": true
 }
@@ -1,9 +0,0 @@
 {
  "id": "patch-1784043007770-e97206c3",
  "patch_name": "gitops-isolated-worktree-governance-20260714T232948UTC8",
  "branch": "patch/gitops-isolated-worktree-governance-20260714T232948UTC8",
  "risk": "high",
  "description": "Root-cause remediation for recurring ACL and shared-worktree failures: unified aiman:gitops runtime identity, UMask=0002, inherited default ACLs, per-task isolated Git worktrees, startup preflight, migration/rollback scripts, systemd drop-in templates, audit evidence, and an OpenClaw execution task. No production business deployment in this patch.",
  "created_by": "chatgpt-gitops-bot",
  "created_at": "2026-07-14T15:30:07Z"
 }
@@ -1,9 +0,0 @@
 {
  "type": "branch",
  "branch": "reconcile/v1r4c5-release-to-main-20260711",
  "base": "main",
  "created_by": "chatgpt-gitops-bot",
  "created_at": "2026-07-10T23:56:41Z",
  "direct_main_push": false,
  "owner_approval_required": true
 }
@@ -1,9 +0,0 @@
 {
  "type": "branch",
  "branch": "task/codex/gitops-isolated-worktree-migration-20260715T001652UTC8",
  "base": "patch/gitops-isolated-worktree-governance-20260714T232948UTC8",
  "created_by": "chatgpt-gitops-bot",
  "created_at": "2026-07-14T16:17:28Z",
  "direct_main_push": false,
  "owner_approval_required": true
 }
@@ -1,9 +0,0 @@
 {
  "id": "patch-1783707996243-33717dfe",
  "patch_name": "v1r4c6-admin-search-agent-integration-20260711",
  "branch": "release/v1r4c6-admin-search-agent-integration-20260711",
  "risk": "high",
  "description": "V1R4C6 bundled integration release: Admin Workbench visibility, SR-03 Search Gateway real provider binding, and Agent Platform Integration. Memory search remains disabled and user data isolation gates remain mandatory.",
  "created_by": "chatgpt-gitops-bot",
  "created_at": "2026-07-10T18:26:36Z"
 }
@@ -0,0 +1,80 @@
 version: 1
 name: owner-maintenance-gate
 baseline: V1R4C3
 created_at: "2026-07-10"
 purpose: Owner-controlled non-dangerous maintenance permissions without arbitrary host CLI access.
 principles:
  - agents_may_detect_and_request
  - owner_may_approve_non_dangerous_actions
  - runner_executes_only_allowlisted_command_keys
  - no_arbitrary_shell
  - protected_branch_merge_is_default_owner_gate
  - every_execution_must_be_audited
 permission_tiers:
  READ_ONLY:
    agent_allowed: true
    owner_allowed: true
    requires_owner_gate: false
  AUTO_PATCH:
    agent_allowed: true
    owner_allowed: true
    requires_owner_gate: false
  OWNER_NON_DANGEROUS_ALL:
    description: Owner may execute all non-destructive allowlisted maintenance command_keys after owner gate approval.
    agent_allowed: false
    owner_allowed: true
    requires_owner_gate: true
    approval_model: protected_branch_merge_or_signed_owner_commit
  OWNER_HIGH_RISK:
    description: main merge, baseline promotion, tag creation, credential/provider mutation, destructive cleanup, production restart, SSH/firewall mutation.
    agent_allowed: false
    owner_allowed: true
    requires_owner_gate: true
    extra_review_required: true
 safety_classification:
  non_dangerous_requirements:
    destructive: false
    arbitrary_shell: false
    credential_mutation: false
    provider_mutation: false
    external_network_write: false
    production_data_delete: false
    baseline_or_tag_mutation: false
    main_branch_direct_write: false
    command_key_must_be_allowlisted: true
  always_high_risk:
    - merge_to_main
    - baseline_promotion
    - tag_creation
    - credential_rotation
    - embedding_provider_change
    - database_delete_or_drop
    - rm_recursive_or_force
    - service_restart_production
    - firewall_or_ssh_config_change
    - arbitrary_bash_or_shell
 allowlist:
  openclaw.memory.index.force:
    permission: OWNER_NON_DANGEROUS_ALL
    risk: medium
    destructive: false
    command_vector:
      - /usr/local/bin/openclaw
      - memory
      - index
      - --force
    timeout_seconds: 1800
    lock_key: memory-index-rebuild
    acceptance:
      - openclaw memory search "V1R4C3"
      - openclaw memory search "Gitea Branch Sync"
      - openclaw memory search "Autonomous Controller"
    success_signals:
      - AIOS_OWNER_MAINTENANCE_MEMORY_INDEX_REBUILT
      - AIOS_MEMORY_SEARCH_RESTORED
 status: proposed
@@ -0,0 +1,133 @@
 #!/usr/bin/env python3
 """Owner maintenance allowlist runner.
 Executes only approved command_key mappings. It never runs arbitrary shell strings.
 """
 from __future__ import annotations
 import argparse
 import hashlib
 import json
 import subprocess
 import sys
 from dataclasses import dataclass
 from datetime import datetime, timezone
 from pathlib import Path
 from typing import Any
 try:
    import yaml  # type: ignore
 except Exception:
    yaml = None
@dataclass(frozen=True)
 class AllowedCommand:
    command_key: str
    argv: tuple[str, ...]
    permission: str
    risk: str
    destructive: bool
    timeout_seconds: int
 ALLOWED_COMMANDS: dict[str, AllowedCommand] = {
    "openclaw.memory.index.force": AllowedCommand(
        command_key="openclaw.memory.index.force",
        argv=("/usr/local/bin/openclaw", "memory", "index", "--force"),
        permission="OWNER_NON_DANGEROUS_ALL",
        risk="medium",
        destructive=False,
        timeout_seconds=1800,
    ),
 }
 DENY_FLAGS = {
    "arbitrary_shell",
    "credential_mutation",
    "provider_mutation",
    "external_network_write",
    "production_data_delete",
    "baseline_or_tag_mutation",
    "main_branch_direct_write",
 }
 def sha256_text(value: str) -> str:
    return hashlib.sha256(value.encode("utf-8", errors="replace")).hexdigest()
 def utc_now() -> str:
    return datetime.now(timezone.utc).isoformat()
 def load_request(path: Path) -> dict[str, Any]:
    if yaml is None:
        raise RuntimeError("PyYAML is required to load owner request YAML")
    data = yaml.safe_load(path.read_text(encoding="utf-8"))
    if not isinstance(data, dict):
        raise ValueError("owner request must be a YAML object")
    return data
 def validate_request(req: dict[str, Any]) -> AllowedCommand:
    command_key = req.get("command_key")
    if command_key not in ALLOWED_COMMANDS:
        raise PermissionError(f"command_key is not allowlisted: {command_key}")
    allowed = ALLOWED_COMMANDS[command_key]
    if req.get("permission") != allowed.permission:
        raise PermissionError("request permission does not match allowlist")
    safety = req.get("safety", {})
    if not isinstance(safety, dict):
        raise PermissionError("safety metadata is required")
    if bool(safety.get("destructive", True)) != allowed.destructive:
        raise PermissionError("destructive flag does not match allowlist")
    denied = [flag for flag in DENY_FLAGS if safety.get(flag) is True]
    if denied:
        raise PermissionError(f"denied safety flags: {denied}")
    gate = req.get("owner_gate", {})
    if not isinstance(gate, dict) or gate.get("required") is not True:
        raise PermissionError("owner_gate.required must be true")
    if not gate.get("approved_by") or not gate.get("approved_at") or not gate.get("approval_commit"):
        raise PermissionError("owner approval fields are required before execution")
    return allowed
 def run_allowed_command(allowed: AllowedCommand, dry_run: bool) -> dict[str, Any]:
    if dry_run:
        return {"dry_run": True, "argv": list(allowed.argv), "exit_code": None}
    proc = subprocess.run(
        list(allowed.argv),
        check=False,
        capture_output=True,
        text=True,
        timeout=allowed.timeout_seconds,
    )
    return {
        "dry_run": False,
        "argv": list(allowed.argv),
        "exit_code": proc.returncode,
        "executed_at": utc_now(),
        "stdout_sha256": sha256_text(proc.stdout),
        "stderr_sha256": sha256_text(proc.stderr),
    }
 def main() -> int:
    parser = argparse.ArgumentParser(description="Run an owner-approved allowlisted maintenance request")
    parser.add_argument("request", type=Path)
    parser.add_argument("--dry-run", action="store_true")
    args = parser.parse_args()
    try:
        req = load_request(args.request)
        allowed = validate_request(req)
        result = run_allowed_command(allowed, dry_run=args.dry_run)
        print(json.dumps({"status": "ok", "command_key": allowed.command_key, "result": result}, indent=2))
        return 0 if result.get("exit_code") in (None, 0) else int(result["exit_code"])
    except Exception as exc:
        print(json.dumps({"status": "error", "error": str(exc)}, indent=2), file=sys.stderr)
        return 1
 if __name__ == "__main__":
    raise SystemExit(main())
@@ -1,47 +0,0 @@
 #!/usr/bin/env bash
 set -euo pipefail
 ROOT="${AIOS_GITOPS_ROOT:-/data/aios-gitops}"
 SOURCE="${AIOS_SOURCE_WORKTREE:-$ROOT/workspace}"
 BARE="${AIOS_BARE_REPO:-$ROOT/repo.git}"
 WORKTREES="${AIOS_WORKTREE_ROOT:-$ROOT/worktrees}"
 RUNTIME_USER="${AIOS_RUNTIME_USER:-aiman}"
 RUNTIME_GROUP="${AIOS_RUNTIME_GROUP:-gitops}"
 if [[ "$(id -u)" -ne 0 ]]; then
  echo "ERROR: bootstrap must run as root" >&2
  exit 2
 fi
 getent passwd "$RUNTIME_USER" >/dev/null
 getent group "$RUNTIME_GROUP" >/dev/null
 [[ -d "$SOURCE/.git" || -f "$SOURCE/.git" ]] || {
  echo "ERROR: source worktree not found: $SOURCE" >&2
  exit 3
 }
 install -d -o "$RUNTIME_USER" -g "$RUNTIME_GROUP" -m 2775 "$ROOT" "$WORKTREES"
 install -d -o "$RUNTIME_USER" -g "$RUNTIME_GROUP" -m 2775 \
  "$WORKTREES/openclaw" "$WORKTREES/chatgpt" "$WORKTREES/controller"
 if [[ ! -d "$BARE" ]]; then
  sudo -u "$RUNTIME_USER" -g "$RUNTIME_GROUP" git clone --bare "$SOURCE" "$BARE"
 fi
 chown -R "$RUNTIME_USER:$RUNTIME_GROUP" "$BARE" "$WORKTREES"
 find "$BARE" "$WORKTREES" -type d -exec chmod 2775 {} +
 find "$BARE" "$WORKTREES" -type f -exec chmod ug+rw {} +
 setfacl -R -m "u:${RUNTIME_USER}:rwX,g:${RUNTIME_GROUP}:rwX,m::rwX" "$BARE" "$WORKTREES"
 find "$BARE" "$WORKTREES" -type d -exec \
  setfacl -m "d:u:${RUNTIME_USER}:rwx,d:g:${RUNTIME_GROUP}:rwx,d:m::rwx" {} +
 cat <<EOF
 bootstrap_status: PASS
 root: $ROOT
 source_worktree: $SOURCE
 bare_repo: $BARE
 worktree_root: $WORKTREES
 runtime_user: $RUNTIME_USER
 runtime_group: $RUNTIME_GROUP
 EOF
@@ -1,48 +0,0 @@
 #!/usr/bin/env bash
 set -euo pipefail
 EXPECTED_USER="${AIOS_RUNTIME_USER:-aiman}"
 EXPECTED_GROUP="${AIOS_RUNTIME_GROUP:-gitops}"
 EXPECTED_UMASK="${AIOS_RUNTIME_UMASK:-0002}"
 TASK_DIR="${AIOS_TASK_DIR:-.agent/tasks}"
 RESULT_DIR="${AIOS_RESULT_DIR:-.agent/results}"
 fail() {
  echo "preflight_status: FAIL"
  echo "reason: $1"
  exit 1
 }
 USER_NOW="$(id -un)"
 GROUPS_NOW="$(id -Gn)"
 UMASK_NOW="$(umask)"
 ROOT="$(git rev-parse --show-toplevel 2>/dev/null || true)"
 GIT_DIR="$(git rev-parse --git-dir 2>/dev/null || true)"
 BRANCH="$(git branch --show-current 2>/dev/null || true)"
 [[ "$USER_NOW" == "$EXPECTED_USER" ]] || fail "runtime user is $USER_NOW, expected $EXPECTED_USER"
 [[ " $GROUPS_NOW " == *" $EXPECTED_GROUP "* ]] || fail "runtime user is not in $EXPECTED_GROUP"
 [[ "$UMASK_NOW" == "$EXPECTED_UMASK" || "$UMASK_NOW" == "002" ]] || fail "umask is $UMASK_NOW, expected $EXPECTED_UMASK"
 [[ -n "$ROOT" ]] || fail "not inside a git worktree"
 [[ "$ROOT" == /data/aios-gitops/worktrees/* ]] || fail "shared source worktree is prohibited: $ROOT"
 [[ -w "$ROOT" ]] || fail "worktree root is not writable"
 [[ -w "$GIT_DIR" ]] || fail "git dir is not writable: $GIT_DIR"
 [[ ! -e "$GIT_DIR/index.lock" ]] || fail "git index lock exists"
 install -d -m 2775 "$TASK_DIR" "$RESULT_DIR"
 [[ -w "$TASK_DIR" ]] || fail "task directory is not writable: $TASK_DIR"
 [[ -w "$RESULT_DIR" ]] || fail "result directory is not writable: $RESULT_DIR"
 [[ -z "$(git status --porcelain)" ]] || fail "worktree is not clean at task start"
 cat <<EOF
 preflight_status: PASS
 runtime_user: $USER_NOW
 runtime_groups: $GROUPS_NOW
 umask: $UMASK_NOW
 worktree: $ROOT
 git_dir: $GIT_DIR
 branch: $BRANCH
 workspace_clean: true
 task_path_writable: true
 result_path_writable: true
 git_lock_present: false
 EOF
@@ -1,73 +0,0 @@
 #!/usr/bin/env bash
 set -euo pipefail
 ACTION="${1:-}"
 AGENT="${2:-}"
 TASK_ID="${3:-}"
 BASE_REF="${4:-origin/main}"
 ROOT="${AIOS_GITOPS_ROOT:-/data/aios-gitops}"
 BARE="${AIOS_BARE_REPO:-$ROOT/repo.git}"
 WORKTREES="${AIOS_WORKTREE_ROOT:-$ROOT/worktrees}"
 RUNTIME_USER="${AIOS_RUNTIME_USER:-aiman}"
 RUNTIME_GROUP="${AIOS_RUNTIME_GROUP:-gitops}"
 usage() {
  echo "usage: $0 create|remove <openclaw|chatgpt|controller> <task-id> [base-ref]" >&2
  exit 64
 }
 [[ "$ACTION" =~ ^(create|remove)$ ]] || usage
 [[ "$AGENT" =~ ^(openclaw|chatgpt|controller)$ ]] || usage
 [[ "$TASK_ID" =~ ^[A-Za-z0-9._-]+$ ]] || {
  echo "ERROR: invalid task id" >&2
  exit 65
 }
 [[ "$(id -un)" == "$RUNTIME_USER" ]] || {
  echo "ERROR: must run as $RUNTIME_USER" >&2
  exit 66
 }
 [[ "$(id -gn)" == "$RUNTIME_GROUP" || " $(id -Gn) " == *" $RUNTIME_GROUP "* ]] || {
  echo "ERROR: user must be in $RUNTIME_GROUP" >&2
  exit 67
 }
 [[ "$(umask)" == "0002" || "$(umask)" == "002" ]] || {
  echo "ERROR: umask must be 0002" >&2
  exit 68
 }
 [[ -d "$BARE" ]] || {
  echo "ERROR: bare repository missing: $BARE" >&2
  exit 69
 }
 PATHNAME="$WORKTREES/$AGENT/$TASK_ID"
 BRANCH="task/$AGENT/$TASK_ID"
 case "$ACTION" in
  create)
    [[ ! -e "$PATHNAME" ]] || {
      echo "ERROR: worktree already exists: $PATHNAME" >&2
      exit 70
    }
    git --git-dir="$BARE" fetch --all --prune
    git --git-dir="$BARE" worktree add -b "$BRANCH" "$PATHNAME" "$BASE_REF"
    install -d -m 2775 "$PATHNAME/.agent/tasks/$AGENT" "$PATHNAME/.agent/results/$AGENT"
    setfacl -m "u:${RUNTIME_USER}:rwx,g:${RUNTIME_GROUP}:rwx,m::rwx" "$PATHNAME"
    find "$PATHNAME" -type d -exec chmod g+s {} +
    find "$PATHNAME" -type d -exec \
      setfacl -m "d:u:${RUNTIME_USER}:rwx,d:g:${RUNTIME_GROUP}:rwx,d:m::rwx" {} +
    printf 'worktree: %s\nbranch: %s\nbase: %s\n' "$PATHNAME" "$BRANCH" "$BASE_REF"
    ;;
  remove)
    [[ -d "$PATHNAME" ]] || {
      echo "ERROR: worktree not found: $PATHNAME" >&2
      exit 71
    }
    [[ -z "$(git -C "$PATHNAME" status --porcelain)" ]] || {
      echo "ERROR: worktree is not clean; retain for investigation" >&2
      exit 72
    }
    git --git-dir="$BARE" worktree remove "$PATHNAME"
    printf 'removed: %s\n' "$PATHNAME"
    ;;
 esac
@@ -1,63 +0,0 @@
 from __future__ import annotations
 import importlib.util
 import json
 import subprocess
 from pathlib import Path
 PATCHER_PATH = Path(__file__).resolve().parents[1] / 'fixes' / 'patch_openclaw_weixin_bigint.py'
 def load_patcher():
    spec = importlib.util.spec_from_file_location('patch_openclaw_weixin_bigint', PATCHER_PATH)
    assert spec is not None and spec.loader is not None
    module = importlib.util.module_from_spec(spec)
    spec.loader.exec_module(module)
    return module
 def test_api_patch_is_idempotent_and_scoped() -> None:
    patcher = load_patcher()
    ts = '\n'.join([
        'import { fileURLToPath } from "node:url";',
        'import { redactBody, redactUrl } from "../util/redact.js";',
        'const resp: GetUpdatesResp = JSON.parse(rawText);',
    ])
    js = '\n'.join([
        'import { fileURLToPath } from "node:url";',
        'import { redactBody, redactUrl } from "../util/redact.js";',
        'const resp = JSON.parse(rawText);',
    ])
    types = 'export interface WeixinMessage {\n  message_id?: number;\n}\n'
    ts_once = patcher.patch_api_ts(ts)
    js_once = patcher.patch_api_js(js)
    types_once = patcher.patch_types_ts(types)
    assert patcher.patch_api_ts(ts_once) == ts_once
    assert patcher.patch_api_js(js_once) == js_once
    assert patcher.patch_types_ts(types_once) == types_once
    assert 'parseWeixinGetUpdatesJson(rawText)' in ts_once
    assert 'parseWeixinGetUpdatesJson(rawText)' in js_once
    assert 'message_id?: string | number;' in types_once
 def test_node_reviver_preserves_only_weixin_exact_ids() -> None:
    script = '''
 const ids = new Set(["message_id", "msg_id", "root_id", "parent_id", "server_msg_id", "new_msg_id"]);
 const parsed = JSON.parse('{"ret":0,"seq":42,"msgs":[{"message_id":7482038281766381234,"create_time_ms":1783856936191,"item_list":[{"msg_id":"3123141975110186863"}]}]}', (key, value, context) => {
  if (!ids.has(key) || value == null || typeof value === "string") return value;
  if (typeof value !== "number" || !context?.source || !/^(?:0|[1-9]\\d*)$/.test(context.source)) throw new Error("unsafe id");
  return context.source;
 });
 console.log(JSON.stringify({ messageId: parsed.msgs[0].message_id, msgId: parsed.msgs[0].item_list[0].msg_id, seqType: typeof parsed.seq, timestampType: typeof parsed.msgs[0].create_time_ms }));
 '''
    completed = subprocess.run(['node', '-e', script], check=True, capture_output=True, text=True)
    result = json.loads(completed.stdout)
    assert result == {
        'messageId': '7482038281766381234',
        'msgId': '3123141975110186863',
        'seqType': 'number',
        'timestampType': 'number',
    }
@@ -1,563 +0,0 @@
 from __future__ import annotations
 import importlib.util
 import json
 import subprocess
 from pathlib import Path
 import pytest
 def load_js_process_helpers() -> str:
    patcher_path = Path(__file__).resolve().parents[1] / 'fixes' / 'patch_openclaw_weixin_hermes.py'
    assert patcher_path.exists(), f'missing patcher: {patcher_path}'
    spec = importlib.util.spec_from_file_location('patch_openclaw_weixin_hermes', patcher_path)
    assert spec is not None and spec.loader is not None
    module = importlib.util.module_from_spec(spec)
    spec.loader.exec_module(module)
    return module.JS_PROCESS_HELPERS
 def load_patcher_module() -> object:
    patcher_path = Path(__file__).resolve().parents[1] / 'fixes' / 'patch_openclaw_weixin_hermes.py'
    assert patcher_path.exists(), f'missing patcher: {patcher_path}'
    spec = importlib.util.spec_from_file_location('patch_openclaw_weixin_hermes', patcher_path)
    assert spec is not None and spec.loader is not None
    module = importlib.util.module_from_spec(spec)
    spec.loader.exec_module(module)
    return module
 def run_js_process_helpers(tmp_path: Path, body: str) -> dict:
    js_process_helpers = load_js_process_helpers()
    node_prelude = '\n'.join(
        [
            'const hermesTestFs = require("node:fs");',
            'globalThis.existsSync = hermesTestFs.existsSync;',
            'globalThis.readFileSync = hermesTestFs.readFileSync;',
            'globalThis.writeFileSync = hermesTestFs.writeFileSync;',
            'globalThis.mkdirSync = hermesTestFs.mkdirSync;',
            'globalThis.appendFileSync = hermesTestFs.appendFileSync;',
            'globalThis.path = require("node:path");',
            'globalThis.logger = { info() {}, warn() {}, error() {} };',
        ]
    )
    node_script = tmp_path / 'hermes_image_followup_check.js'
    node_script.write_text(node_prelude + '\n\n' + js_process_helpers + '\n\n' + body, encoding='utf-8')
    completed = subprocess.run(['node', str(node_script)], check=True, capture_output=True, text=True)
    return json.loads(completed.stdout.strip())
@pytest.mark.parametrize('is_ts', [True, False], ids=['ts', 'js'])
 def test_replace_or_insert_process_helpers_is_byte_idempotent(is_ts: bool) -> None:
    patcher = load_patcher_module()
    marker = 'export async function processOneMessage'
    source = marker + '(message) {\n  return message;\n}\n'
    once = patcher.replace_or_insert_process_helpers(source, is_ts=is_ts)
    twice = patcher.replace_or_insert_process_helpers(once, is_ts=is_ts)
    helpers = patcher.TS_PROCESS_HELPERS if is_ts else patcher.JS_PROCESS_HELPERS
    assert helpers + '\n\n' + marker in once
    assert twice == once
@pytest.mark.parametrize(
    ('is_ts', 'hook_name'),
    [(True, 'TS_PROCESS_HOOK'), (False, 'JS_PROCESS_HOOK')],
    ids=['ts', 'js'],
 )
 def test_remove_and_reinsert_process_hook_is_byte_idempotent(is_ts: bool, hook_name: str) -> None:
    patcher = load_patcher_module()
    hook = getattr(patcher, hook_name)
    indent = '  ' if is_ts else '    '
    needle = indent + 'ctx.CommandAuthorized = commandAuthorized;\n'
    next_nonblank = indent + 'continueProcessing();\n'
    def cycle(text: str) -> str:
        without_hook = patcher.remove_existing_process_hermes_hook(text)
        return without_hook.replace(needle, needle + hook + '\n', 1)
    once = cycle(needle + next_nonblank)
    twice = cycle(once)
    assert hook + '\n' + next_nonblank in once
    assert twice == once
 def test_looks_like_hermes_image_followup_distinguishes_diagnostics_from_real_image_questions(tmp_path: Path) -> None:
    cases = {
        'diag_cache_expired': ('\u56fe\u7247\u7f13\u5b58\u5df2\u5931\u6548\uff0c\u8bf7\u91cd\u65b0\u53d1\u9001\u56fe\u7247\u3002', False),
        'diag_cache_error_en': ('image cache error', False),
        'diag_cache_expired_en': ('the image cache expired, please resend', False),
        'diag_copying_text_shows_cache_error': ('\u4e3a\u4ec0\u4e48\u6211\u590d\u5236\u53d1\u7ed9\u4f60\u7684\u6587\u5b57\uff0c\u4f60\u90fd\u663e\u793a\u56fe\u7247\u7f13\u5b58\u5df2\u5931\u6548\uff0c\u8bf7\u91cd\u65b0\u53d1\u9001\u56fe\u7247\u3002', False),
        'diag_not_image': ('\u8fd9\u4e0d\u662f\u56fe\u7247\uff0c\u662f\u6587\u5b57\u6d88\u606f', False),
        'diag_text_not_image_en': ('this is a text message, not an image', False),
        'diag_no_image_sent': ('\u6211\u6ca1\u6709\u53d1\u56fe\u7247', False),
        'diag_system_logs': ('\u8bf7\u68c0\u67e5\u7cfb\u7edf\u65e5\u5fd7\uff0c\u4ece\u673a\u5236\u4e0a\u4fee\u590d\u9519\u8bef', False),
        'image_cn_1': ('\u8bf7\u770b\u4e00\u4e0b\u8fd9\u5f20\u56fe\u7247', True),
        'image_cn_2': ('\u8fd9\u56fe\u662f\u4ec0\u4e48', True),
        'image_cn_3': ('\u8bc6\u522b\u56fe\u7247\u5185\u5bb9', True),
        'image_en': ('look at this image', True),
    }
    results = run_js_process_helpers(
        tmp_path,
        'const cases = '
        + json.dumps({key: text for key, (text, _) in cases.items()}, ensure_ascii=False)
        + ';\nconst results = {};\n'
        + 'for (const [key, text] of Object.entries(cases)) {\n'
        + '  results[key] = looksLikeHermesImageFollowup(text);\n'
        + '}\n'
        + 'console.log(JSON.stringify(results));\n',
    )
    for key, (_, expected) in cases.items():
        assert results[key] is expected, f'{key}: expected {expected}, got {results[key]!r}'
 def test_resolve_hermes_referenced_image_followup_requires_exact_cache(tmp_path: Path) -> None:
    cache_path = tmp_path / 'hermes-last-image-cache.json'
    results = run_js_process_helpers(
        tmp_path,
        '\n'.join(
            [
                'const fs = require("fs");',
                f'const cachePath = {json.dumps(str(cache_path))};',
                'const accountId = "account-a";',
                'const senderId = "sender-b";',
                'const exactQuotedId = "quoted-123";',
                'const cache = {};',
                'cache[hermesLastImageCacheKey(accountId, senderId)] = {',
                '  mediaUrl: "data:image/png;base64,iVBORw0KGgo=",',
                '  timestampMs: Date.now(),',
                '  messageId: "latest-999",',
                '  mediaType: "image/png",',
                '};',
                'cache[hermesExactImageCacheKey(accountId, senderId, exactQuotedId)] = {',
                '  mediaUrl: "data:image/png;base64,iVBORw0KGgo=",',
                '  timestampMs: Date.now(),',
                '  messageId: exactQuotedId,',
                '  mediaType: "image/png",',
                '};',
                'fs.writeFileSync(cachePath, JSON.stringify(cache, null, 2), "utf-8");',
                'process.env.AIOS_HERMES_WEIXIN_LAST_IMAGE_CACHE = cachePath;',
                'const generic = resolveHermesReferencedImageFollowup(accountId, senderId, "plain quoted text", "quoted-999");',
                'const exact = resolveHermesReferencedImageFollowup(accountId, senderId, "plain quoted text", exactQuotedId);',
                'const forced = resolveHermesFollowupImageContext(accountId, senderId, "plain quoted text", exactQuotedId, false, true);',
                'console.log(JSON.stringify({ generic, exact, forced }));',
            ]
        ),
    )
    assert results['generic']['imageIntent'] is False
    assert results['generic']['matched'] is False
    assert results['generic']['reason'] == 'referenced_image_not_confirmed'
    assert results['exact']['imageIntent'] is True
    assert results['exact']['matched'] is True
    assert results['exact']['reason'] == 'cached_image_matched'
    assert results['forced']['imageIntent'] is True
    assert results['forced']['matched'] is True
 def test_referenced_text_without_media_evidence_stays_out_of_image_cache(tmp_path: Path) -> None:
    cache_path = tmp_path / 'hermes-last-image-cache.json'
    results = run_js_process_helpers(
        tmp_path,
        '\n'.join(
            [
                'const fs = require("fs");',
                f'const cachePath = {json.dumps(str(cache_path))};',
                'const accountId = "account-a";',
                'const senderId = "sender-b";',
                'const cache = {};',
                'cache[hermesLastImageCacheKey(accountId, senderId)] = {',
                '  mediaUrl: "data:image/png;base64,iVBORw0KGgo=",',
                '  timestampMs: Date.now(),',
                '  messageId: "latest-999",',
                '  mediaType: "image/png",',
                '};',
                'fs.writeFileSync(cachePath, JSON.stringify(cache, null, 2), "utf-8");',
                'process.env.AIOS_HERMES_WEIXIN_LAST_IMAGE_CACHE = cachePath;',
                'const full = { item_list: [{ ref_msg: { message_item: { type: 0, id: "quoted-999" } } }] };',
                'const hermesContent = "\u8fd9\u662f\u4ec0\u4e48";',
                'const hermesReferenceContextPresent = hasHermesReferenceContext(full.item_list);',
                'const hermesQuotedMessageId = findHermesReferencedMessageId(full.item_list);',
                'const refMediaItem = undefined;',
                'const hermesReferencedMediaIsImage = Boolean(refMediaItem && hermesCurrentMessageIsImage({}, refMediaItem));',
                'const hermesReferencedImageFollowupIntent = hermesReferenceContextPresent && !refMediaItem;',
                'const hermesCurrentImage = false;',
                'const hermesFollowup = hermesCurrentImage',
                '  ? { imageIntent: false, matched: false, reason: "current_image_message", cacheStrategy: "current", quotedMessageId: hermesQuotedMessageId }',
                '  : (hermesReferencedMediaIsImage',
                '      ? resolveHermesFollowupImageContext(accountId, senderId, hermesContent, hermesQuotedMessageId, false, true)',
                '      : (hermesReferencedImageFollowupIntent',
                '          ? resolveHermesReferencedImageFollowup(accountId, senderId, hermesContent, hermesQuotedMessageId)',
                '          : resolveHermesFollowupImageContext(accountId, senderId, hermesContent)));',
                'console.log(JSON.stringify({ hermesReferenceContextPresent, hermesQuotedMessageId, hermesReferencedMediaIsImage, hermesReferencedImageFollowupIntent, hermesFollowup }));',
            ]
        ),
    )
    assert results['hermesReferenceContextPresent'] is True
    assert results['hermesQuotedMessageId'] == 'quoted-999'
    assert results['hermesReferencedMediaIsImage'] is False
    assert results['hermesReferencedImageFollowupIntent'] is True
    assert results['hermesFollowup']['imageIntent'] is False
    assert results['hermesFollowup']['matched'] is False
    assert results['hermesFollowup']['reason'] == 'referenced_image_not_confirmed'
 def test_explicit_quoted_text_builds_grounded_plain_text_context(tmp_path: Path) -> None:
    results = run_js_process_helpers(
        tmp_path,
        '\n'.join(
            [
                'const available = typeof resolveHermesQuotedTextContext === "function";',
                'const items = [{ type: 1, ref_msg: { message_item: { type: 1, text_item: { text: "被引用的正文" } } } }];',
                'const result = available ? resolveHermesQuotedTextContext(items, "这是什么意思？") : null;',
                'console.log(JSON.stringify({ available, result }));',
            ]
        ),
    )
    assert results['available'] is True
    assert results['result'] == {
        'referenced': True,
        'handled': False,
        'prompt': '[被引用文字]\n被引用的正文\n[/被引用文字]\n\n[当前问题]\n这是什么意思？\n[/当前问题]',
        'reason': 'quoted_text_resolved',
    }
 def test_visible_ref_title_builds_grounded_plain_text_context(tmp_path: Path) -> None:
    results = run_js_process_helpers(
        tmp_path,
        '\n'.join(
            [
                'const items = [{ type: 1, ref_msg: { title: "客户端可见的引用正文", message_item: { type: 0, msg_id: "quoted-123" } } }];',
                'const result = resolveHermesQuotedTextContext(items, "这是什么意思？");',
                'console.log(JSON.stringify(result));',
            ]
        ),
    )
    assert results == {
        'referenced': True,
        'handled': False,
        'prompt': '[被引用文字]\n客户端可见的引用正文\n[/被引用文字]\n\n[当前问题]\n这是什么意思？\n[/当前问题]',
        'reason': 'quoted_text_resolved',
    }
 def test_nested_generic_title_is_not_treated_as_quoted_text(tmp_path: Path) -> None:
    results = run_js_process_helpers(
        tmp_path,
        '\n'.join(
            [
                'const items = [{ type: 1, ref_msg: { message_item: { type: 0, title: "非协议引用摘要" } } }];',
                'const result = resolveHermesQuotedTextContext(items, "这是什么意思？");',
                'console.log(JSON.stringify(result));',
            ]
        ),
    )
    assert results == {
        'referenced': True,
        'handled': True,
        'replyText': '无法读取被引用文字，请复制原文后再提问。',
        'reason': 'quoted_text_unavailable',
    }
 def test_exact_message_id_cache_resolves_server_reference_without_guessing(tmp_path: Path) -> None:
    cache_path = tmp_path / 'hermes-message-text-cache.json'
    results = run_js_process_helpers(
        tmp_path,
        '\n'.join(
            [
                f'process.env.AIOS_HERMES_WEIXIN_TEXT_MESSAGE_CACHE = {json.dumps(str(cache_path))};',
                'const accountId = "account-a";',
                'const senderId = "sender-b";',
                'const original = { item_list: [{ type: 1, msg_id: "7481729105240418312", text_item: { text: "客户端已经显示的原始正文" } }] };',
                'rememberHermesTextMessage(accountId, senderId, original, "客户端已经显示的原始正文");',
                'const quotedItems = [{ type: 1, ref_msg: { message_item: { type: 0, msg_id: "7481729105240418312" } }, text_item: { text: "这是什么意思？" } }];',
                'const quotedId = findHermesReferencedMessageId(quotedItems);',
                'const exact = resolveHermesCachedQuotedText(accountId, senderId, quotedId);',
                'const wrong = resolveHermesCachedQuotedText(accountId, senderId, "7481729105240418313");',
                'const resolved = resolveHermesQuotedTextContext(quotedItems, "这是什么意思？", exact);',
                'const mode = (hermesTestFs.statSync(process.env.AIOS_HERMES_WEIXIN_TEXT_MESSAGE_CACHE).mode & 0o777).toString(8);',
                'console.log(JSON.stringify({ quotedId, exact, wrong, resolved, mode }));',
            ]
        ),
    )
    assert results['quotedId'] == '7481729105240418312'
    assert results['exact'] == '客户端已经显示的原始正文'
    assert results['wrong'] == ''
    assert results['resolved'] == {
        'referenced': True,
        'handled': False,
        'prompt': '[被引用文字]\n客户端已经显示的原始正文\n[/被引用文字]\n\n[当前问题]\n这是什么意思？\n[/当前问题]',
        'reason': 'quoted_text_resolved',
    }
    assert results['mode'] == '600'
 def test_expired_message_id_cache_entry_is_not_resolved(tmp_path: Path) -> None:
    cache_path = tmp_path / 'hermes-message-text-cache.json'
    results = run_js_process_helpers(
        tmp_path,
        '\n'.join(
            [
                f'process.env.AIOS_HERMES_WEIXIN_TEXT_MESSAGE_CACHE = {json.dumps(str(cache_path))};',
                'const key = hermesTextMessageCacheKey("account-a", "sender-b", "quoted-old");',
                'hermesTestFs.writeFileSync(process.env.AIOS_HERMES_WEIXIN_TEXT_MESSAGE_CACHE, JSON.stringify({ [key]: { text: "过期正文", timestampMs: Date.now() - (8 * 24 * 60 * 60 * 1000) } }), { encoding: "utf-8", mode: 0o600 });',
                'const result = resolveHermesCachedQuotedText("account-a", "sender-b", "quoted-old");',
                'console.log(JSON.stringify({ result }));',
            ]
        ),
    )
    assert results['result'] == ''
 def test_unsafe_numeric_message_id_is_rejected_before_cache_keying(tmp_path: Path) -> None:
    results = run_js_process_helpers(
        tmp_path,
        '\n'.join(
            [
                'const exactString = extractHermesMessageId({ message_id: "7482038281766381234" });',
                'const unsafeNumber = extractHermesMessageId({ message_id: 7482038281766381234 });',
                'const safeNumber = extractHermesMessageId({ message_id: 12345 });',
                'console.log(JSON.stringify({ exactString, unsafeNumber: unsafeNumber ?? null, safeNumber }));',
            ]
        ),
    )
    assert results == {
        'exactString': '7482038281766381234',
        'unsafeNumber': None,
        'safeNumber': '12345',
    }
 def test_exact_outbound_create_time_resolves_quoted_bot_text(tmp_path: Path) -> None:
    cache_path = tmp_path / 'hermes-message-text-cache.json'
    results = run_js_process_helpers(
        tmp_path,
        '\n'.join(
            [
                f'process.env.AIOS_HERMES_WEIXIN_TEXT_MESSAGE_CACHE = {json.dumps(str(cache_path))};',
                'const accountId = "account-a";',
                'const senderId = "sender-b";',
                'const createTimeMs = 1783857900123;',
                'rememberHermesOutboundTextMessage(accountId, senderId, "机器人出站正文", createTimeMs);',
                'const quotedItems = [{ type: 1, ref_msg: { message_item: { type: 0, msg_id: "server-quoted", create_time_ms: createTimeMs } }, text_item: { text: "进展汇报" } }];',
                'const referencedTime = findHermesReferencedCreateTimeMs(quotedItems);',
                'const exact = resolveHermesCachedQuotedText(accountId, senderId, "server-quoted", referencedTime);',
                'const wrong = resolveHermesCachedQuotedText(accountId, senderId, "server-quoted", createTimeMs + 1);',
                'const resolved = resolveHermesQuotedTextContext(quotedItems, "进展汇报", exact);',
                'console.log(JSON.stringify({ referencedTime, exact, wrong, resolved }));',
            ]
        ),
    )
    assert results['referencedTime'] == '1783857900123'
    assert results['exact'] == '机器人出站正文'
    assert results['wrong'] == ''
    assert results['resolved']['reason'] == 'quoted_text_resolved'
    assert '机器人出站正文' in results['resolved']['prompt']
 def test_outbound_create_time_collision_fails_closed(tmp_path: Path) -> None:
    cache_path = tmp_path / 'hermes-message-text-cache.json'
    results = run_js_process_helpers(
        tmp_path,
        '\n'.join(
            [
                f'process.env.AIOS_HERMES_WEIXIN_TEXT_MESSAGE_CACHE = {json.dumps(str(cache_path))};',
                'const createTimeMs = 1783857900123;',
                'rememberHermesOutboundTextMessage("account-a", "sender-b", "第一条", createTimeMs);',
                'rememberHermesOutboundTextMessage("account-a", "sender-b", "第二条", createTimeMs);',
                'const resolved = resolveHermesCachedQuotedText("account-a", "sender-b", undefined, createTimeMs);',
                'console.log(JSON.stringify({ resolved }));',
            ]
        ),
    )
    assert results['resolved'] == ''
 def test_missing_quoted_text_fails_closed_without_model_prompt(tmp_path: Path) -> None:
    results = run_js_process_helpers(
        tmp_path,
        '\n'.join(
            [
                'const available = typeof resolveHermesQuotedTextContext === "function";',
                'const items = [{ type: 1, ref_msg: { message_item: { type: 0, msg_id: "7481729105240418312" } } }];',
                'const result = available ? resolveHermesQuotedTextContext(items, "这是什么意思？") : null;',
                'let modelCalls = 0;',
                'if (result && !result.handled) modelCalls += 1;',
                'console.log(JSON.stringify({ available, result, modelCalls }));',
            ]
        ),
    )
    assert results['available'] is True
    assert results['result'] == {
        'referenced': True,
        'handled': True,
        'replyText': '无法读取被引用文字，请复制原文后再提问。',
        'reason': 'quoted_text_unavailable',
    }
    assert results['modelCalls'] == 0
@pytest.mark.parametrize('hook_name', ['TS_PROCESS_HOOK', 'JS_PROCESS_HOOK'])
 def test_process_hook_stops_unavailable_quote_before_model_dispatch(hook_name: str) -> None:
    patcher = load_patcher_module()
    hook = getattr(patcher, hook_name)
    guard_name = 'TS_QUOTED_TEXT_DISPATCH_GUARD' if hook_name.startswith('TS_') else 'JS_QUOTED_TEXT_DISPATCH_GUARD'
    guard = getattr(patcher, guard_name)
    assert 'rememberHermesTextMessage(deps.accountId, hermesSenderId, full, textBody)' in hook
    assert 'findHermesReferencedCreateTimeMs(full.item_list)' in hook
    assert 'resolveHermesCachedQuotedText(deps.accountId, hermesSenderId, hermesQuotedMessageId, hermesQuotedCreateTimeMs)' in hook
    assert 'resolveHermesQuotedTextContext(full.item_list, hermesContent, hermesCachedQuotedText)' in hook
    assert '!hermesQuotedTextContext.referenced && !textBody.startsWith("/")' in hook
    assert 'ctx.Body = hermesQuotedTextContext.prompt' in hook
    assert 'await dispatchHermesQuotedTextFailClosed(hermesQuotedTextContext, dispatcher)' in guard
    assert 'markDispatchIdle();' in guard
 def test_dispatch_quoted_text_fail_closed_uses_dispatcher_without_model(tmp_path: Path) -> None:
    result = run_js_process_helpers(
        tmp_path,
        '\n'.join(
            [
                '(async () => {',
                '  const calls = [];',
                '  const dispatcher = {',
                '    sendFinalReply(payload) { calls.push(["final", payload]); },',
                '    markComplete() { calls.push(["complete"]); },',
                '    async waitForIdle() { calls.push(["idle"]); },',
                '  };',
                '  const context = resolveHermesQuotedTextContext([{ ref_msg: { message_item: { type: 0, msg_id: "quoted-1" } } }], "这是什么意思？");',
                '  let modelCalls = 0;',
                '  const handled = await dispatchHermesQuotedTextFailClosed(context, dispatcher);',
                '  if (!handled) modelCalls += 1;',
                '  console.log(JSON.stringify({ handled, modelCalls, calls }));',
                '})();',
            ]
        ),
    )
    assert result['handled'] is True
    assert result['modelCalls'] == 0
    assert result['calls'] == [
        ['final', {'text': '无法读取被引用文字，请复制原文后再提问。'}],
        ['complete'],
        ['idle'],
    ]
@pytest.mark.parametrize(
    ('is_ts', 'guard_name', 'needle'),
    [
        (True, 'TS_QUOTED_TEXT_DISPATCH_GUARD', '  logger.debug(`dispatchReplyFromConfig: starting agentId=${route.agentId ?? "(none)"}`);\n'),
        (False, 'JS_QUOTED_TEXT_DISPATCH_GUARD', '    logger.debug(`dispatchReplyFromConfig: starting agentId=${route.agentId ?? "(none)"}`);\n'),
    ],
    ids=['ts', 'js'],
 )
 def test_dispatch_guard_remove_and_reinsert_is_byte_idempotent(is_ts: bool, guard_name: str, needle: str) -> None:
    patcher = load_patcher_module()
    guard = getattr(patcher, guard_name)
    def cycle(text: str) -> str:
        without = patcher.remove_existing_process_hermes_hook(text)
        return without.replace(needle, guard + '\n' + needle, 1)
    once = cycle(needle)
    twice = cycle(once)
    assert guard + '\n' + needle in once
    assert twice == once
 def test_reference_metadata_content_is_not_treated_as_quoted_text(tmp_path: Path) -> None:
    result = run_js_process_helpers(
        tmp_path,
        '\n'.join(
            [
                'const items = [{ ref_msg: { message_item: { type: 0, msg_id: "quoted-1", content: "metadata only" } } }];',
                'console.log(JSON.stringify(resolveHermesQuotedTextContext(items, "这是什么意思？")));',
            ]
        ),
    )
    assert result['handled'] is True
    assert result['reason'] == 'quoted_text_unavailable'
    assert result['replyText'] == '无法读取被引用文字，请复制原文后再提问。'
 def test_patch_plugin_process_only_skips_slash_and_channel(monkeypatch: pytest.MonkeyPatch, tmp_path: Path) -> None:
    patcher = load_patcher_module()
    calls: list[tuple[str, Path]] = []
    def record(name: str, result):
        def _inner(root: Path):
            calls.append((name, root))
            return result
        return _inner
    monkeypatch.setattr(patcher, 'patch_ts', record('patch_ts', False))
    monkeypatch.setattr(patcher, 'patch_js', record('patch_js', False))
    monkeypatch.setattr(patcher, 'patch_process_messages', record('patch_process_messages', {'src/messaging/process-message.ts': True, 'dist/src/messaging/process-message.js': False}))
    monkeypatch.setattr(patcher, 'patch_channel_files', record('patch_channel_files', {'src/channel.ts': False, 'dist/src/channel.js': False}))
    monkeypatch.setattr(patcher, 'verify_plugin_hook', record('verify_plugin_hook', {
        'hook_status': {},
        'hook_installed': False,
        'process_status': {'src/messaging/process-message.ts': True, 'dist/src/messaging/process-message.js': True},
        'process_hook_installed': True,
        'channel_status': {},
        'channel_hook_installed': False,
        'dispatch_reference_found': False,
        'dispatch_reference_files': [],
    }))
    result = patcher.patch_plugin(tmp_path, process_only=True)
    assert calls == [
        ('patch_process_messages', tmp_path),
        ('verify_plugin_hook', tmp_path),
    ]
    assert result['mode'] == 'process-only'
    assert result['plugin_root'] == str(tmp_path)
    assert result['slash_changes'] == {}
    assert result['channel_changes'] == {}
    assert result['process_changes'] == {'src/messaging/process-message.ts': True, 'dist/src/messaging/process-message.js': False}
    assert result['behavior_changes'] == {'src/messaging/process-message.ts': True, 'dist/src/messaging/process-message.js': False}
    assert result['verification']['process_hook_installed'] is True
 def test_parse_args_supports_default_root_and_process_only_flag() -> None:
    patcher = load_patcher_module()
    defaults = patcher.parse_args([])
    assert defaults.plugin_root == patcher.DEFAULT_PLUGIN_ROOT
    assert defaults.process_only is False
    process_only = patcher.parse_args(['--process-only'])
    assert process_only.plugin_root == patcher.DEFAULT_PLUGIN_ROOT
    assert process_only.process_only is True
    custom_root = Path('/tmp/custom-plugin-root')
    custom = patcher.parse_args([str(custom_root), '--process-only'])
    assert custom.plugin_root == custom_root
    assert custom.process_only is True
@@ -1,75 +0,0 @@
 from __future__ import annotations
 import importlib.util
 from pathlib import Path
 PATCHER_PATH = Path(__file__).resolve().parents[1] / 'fixes' / 'patch_openclaw_weixin_outbound_quotes.py'
 def load_patcher():
    spec = importlib.util.spec_from_file_location('patch_openclaw_weixin_outbound_quotes', PATCHER_PATH)
    assert spec is not None and spec.loader is not None
    module = importlib.util.module_from_spec(spec)
    spec.loader.exec_module(module)
    return module
 def test_send_patch_adds_exact_create_time_and_is_idempotent() -> None:
    patcher = load_patcher()
    ts = '''function buildTextMessageReq(params: {
  to: string;
  text: string;
  contextToken?: string;
  runId?: string;
  clientId: string;
 }): SendMessageReq {
  const { to, text, contextToken, runId, clientId } = params;
  return { msg: { client_id: clientId, message_type: MessageType.BOT } };
 }
 export async function sendMessageWeixin(params: {}): Promise<{ messageId: string }> {
  const { to, text, opts } = params;
  const clientId = generateClientId();
  const req = buildSendMessageReq({
    payload: {},
    clientId,
  });
  try {
    await sendMessageApi({ body: req });
  } catch (err) {
    throw err;
  }
  return { messageId: clientId };
 }
 /** Send a single structured MessageItem downstream. */'''
    js = '''function buildTextMessageReq(params) {
    const { to, text, contextToken, runId, clientId } = params;
    return { msg: { client_id: clientId, message_type: MessageType.BOT } };
 }
 export async function sendMessageWeixin(params) {
    const clientId = generateClientId();
    const req = buildSendMessageReq({
        payload: {},
        clientId,
    });
    try {
        await sendMessageApi({ body: req });
    }
    catch (err) {
        throw err;
    }
    return { messageId: clientId };
 }
 /** Send a single structured MessageItem downstream. */'''
    ts_once = patcher.patch_send_ts(ts)
    js_once = patcher.patch_send_js(js)
    assert patcher.patch_send_ts(ts_once) == ts_once
    assert patcher.patch_send_js(js_once) == js_once
    for result in (ts_once, js_once):
        assert 'const createTimeMs = Date.now();' in result
        assert 'text item missing' in result
        assert 'req.msg.item_list[0].create_time_ms = createTimeMs' in result
        assert 'req.msg.create_time_ms = createTimeMs' not in result
        assert 'return { messageId: clientId, createTimeMs }' in result
@@ -1,38 +0,0 @@
 {
    "version": "V1R4C5",
    "codename": "C5 reliability-security-admin",
    "status": "active-baseline",
    "tag": "V1R4C5",
    "tag_ref": "refs/tags/V1R4C5",
    "tag_object": "ac9bdacb19ff66cba8046d75358866f0f0fffb82",
    "previous_version": "V1R4C3",
    "rollback_version": "V1R4C3",
    "baseline": "V1R4C3",
    "release_reason": "reliability, watcher cleanup, user data isolation gate, and admin workbench read-only preparation; memory search disabled",
    "purpose": "stabilize autonomous operations, enforce user isolation gate, prepare admin workbench read-only for C6",
    "changes": [
        "reliability: idle container GC with nixos-activate fallback retention policy and CI reset",
        "watcher: remove redundant local async watcher, consolidate to server-pushed events",
        "security: user data isolation gate (read-only admin panel, fb isolation check)",
        "admin workbench: read-only preparation layout for V1R4C6 admin-search integration",
        "governance: final owner approval via Owner Gate Web UI"
    ],
    "artifacts": [
        "VERSION",
        "docs/releases/V1R4C5-acceptance-evidence.md",
        "docs/releases/V1R4C5-final-owner-approval.md"
    ],
    "validated_actions": [],
    "secret_plaintext_written": false,
    "owner_approval_required": true,
    "owner_notification_channels": ["wechat", "web_ui_audit"],
    "constraints_original": {
        "memory_search": "remain disabled",
        "sr04": "disable read-only visibility slice if RBAC or isolation check fails",
        "gitea": "enter autonomous safe mode if branch consistency drift detected"
    },
    "promoted_at": "2026-07-11T02:17:31+08:00",
    "promoted_by": "openclaw/admin",
    "created_at": "2026-07-11T01:40:00+08:00",
    "author": "openclaw"
 }
@@ -1,111 +0,0 @@
 {
  "version": "V1R4C6",
  "codename": "admin-search-agent-integration",
  "status": "release-candidate-started",
  "base_baseline": "V1R4C5",
  "created_at": "2026-07-11",
  "created_by": "chatgpt-gitops-bot",
  "release_type": "controlled-bundled-integration",
  "promotion_policy": "owner-gated",
  "owner_decision": {
    "accepted_scope": true,
    "requested_delivery": "admin workbench, SR-03, and Agent Platform Integration in one release train",
    "memory_search_enabled": false,
    "force_push_allowed": false
  },
  "scope": {
    "included_groups": [
      {
        "id": "V1R4C6-A",
        "name": "Admin Workbench and Portal AIOS Read-Only Rollout",
        "branches": [
          "feature/v1r4c4-sr04-admin-workbench-visibility-20260711",
          "feature/v1r4c4-sr04-portal-aios-visibility-20260710",
          "patch/user-portal-me-route-diagnostics-20260709",
          "patch/user-portal-rbac-ops-dashboard",
          "patch/agent-observability-webui",
          "patch/web-ui-complete-gitops-dashboard-20260706"
        ],
        "risk": "medium",
        "delivery": "administrator-visible task progress, process diagrams, health, agent observability, and portal AIOS entry points",
        "guardrails": [
          "read-only first",
          "RBAC enforced",
          "empty states explicit",
          "no cross-user or cross-project data exposure"
        ]
      },
      {
        "id": "V1R4C6-B",
        "name": "SR-03 Search Gateway Real Provider Binding",
        "branches": [
          "feature/v1r4c4-sr02-search-gateway-resilience-20260710",
          "patch/search-gateway-governed-general-capability",
          "patch/gitops-change-archive-audit"
        ],
        "risk": "high",
        "delivery": "real provider binding, governed search gateway, resilience, and audit trail",
        "guardrails": [
          "user data isolation gate required",
          "scope filter before ranking",
          "provenance required",
          "no global memory/vector index"
        ]
      },
      {
        "id": "V1R4C6-C",
        "name": "Agent Platform Integration",
        "branches": [
          "patch/agent-event-bus-unified",
          "patch/agent-gpt-message-plane",
          "patch/autonomous-controller-dispatch-20260709",
          "patch/self-evolution-agent-orchestration-policy",
          "patch/aios-autonomous-orchestrator-architecture-review-v1"
        ],
        "risk": "high",
        "delivery": "agent event bus, GPT message plane, dispatch, orchestration policy, and architecture record",
        "guardrails": [
          "controller safe-mode respected",
          "event/message scopes enforced",
          "owner gate for consequential operations",
          "observability before autonomous expansion"
        ]
      }
    ],
    "excluded": [
      "memory search activation",
      "global vector index",
      "full cross-session recall",
      "unscoped search over memory",
      "bulk merge of test/smoke/auth branches",
      "Gitea DB mutation",
      "force-push reconciliation"
    ]
  },
  "acceptance_required": [
    "getRuntimeStatus active",
    "getBaseline reports V1R4C5 before integration and V1R4C6 after promotion",
    "Admin workbench RBAC read-only check passes",
    "Portal AIOS manifest/dashboard/health routes verified",
    "Flow/process diagram renders or returns explicit empty state",
    "Search gateway provider binding returns scoped, provenanced results only",
    "User data isolation acceptance remains green",
    "Agent event bus and GPT message plane tests pass",
    "Autonomous controller dispatch respects safe mode",
    "Memory search remains disabled"
  ],
  "deployment_constraints": {
    "AIOS_MEMORY_ENABLED": false,
    "memory_search_activation_requires_future_owner_approval": true,
    "force_push_allowed": false,
    "gitea_db_mutation_requires_owner_admin_approval": true,
    "admin_workbench_mode": "read_only_first",
    "search_provider_scope": "scoped_only"
  },
  "rollback_policy": {
    "admin_workbench": "disable new menu/routes if RBAC or route acceptance fails",
    "search_gateway": "fallback to disabled provider binding, not memory search",
    "agent_platform": "disable dispatch expansion and keep event/message plane read-only if safety tests fail",
    "memory_search": "remain disabled"
  }
 }
Author	SHA1	Message	Date
openclaw	68eb88cf04	docs(owner-gate): document owner maintenance gate	2026-07-10 18:33:05 +08:00
openclaw	3218b89a2c	feat(owner-gate): add allowlist-only maintenance runner	2026-07-10 18:32:59 +08:00
openclaw	6ac62c265b	chore(owner-gate): request memory index rebuild via command key	2026-07-10 18:32:48 +08:00
openclaw	2322120f39	feat(owner-gate): add owner maintenance permission policy	2026-07-10 18:32:43 +08:00
openclaw	c27b91c708	chore(owner-gate): create clean owner maintenance gate v2 branch	2026-07-10 18:32:35 +08:00