# G0 Evidence Coverage V2

R2 corrects the R1 wording: WP0a–d and WP1 outputs are **discovery evidence**,
not complete Generation-G0 acceptance by themselves.

| WP | R2 output | What it proves | Still required for G0 Exit |
|---|---|---|---|
| WP0a | PARTIAL_DISCOVERY | authority/runtime/production OID relation | VERSION/current/baseline/README/versions/release metadata reader-writer-caller/live mapping |
| WP0b | PARTIAL_DISCOVERY | lexical Git-writer candidates | writer→auth→repo→ref→CAS→receipt→caller graph and live classification |
| WP0c | PARTIAL_DISCOVERY | authorization references by term/subdomain hint | owner/broker, TaskPrincipal, ClaimLease/TaskLease, WriterLease, release/recovery, signer chain + revocation/live caller |
| WP0d | PARTIAL_DISCOVERY | function-level canonical/digest candidates | producer/consumer, input schema, bytes rule, algorithm and duplicate semantic groups |
| WP0e | caller_scan v2 | fail-closed source/runtime caller evidence | five historical known-case acceptance + complete runtime evidence for any PROVEN_ORPHAN |
| WP1 | capability discovery / requirements evaluated | actual local capability probes | declared host requirements for each key execution host, missing requirements resolved |
| WP2 | conservative decision layer | retry/suppress/reconcile/block decision | audit baseline on real operations; no scheduler authority |

`g0_exit_eligible=false` in WP0a–d output is intentional. The G0 gate is a
separate acceptance aggregation step and must reference the missing evidence
listed above.

WP3–WP10 remain dependency-gated backlog exactly as Platform Baseline V4
defines; R2 does not silently redefine the long-term platform goal to U01–U09.