062-review-verdict-status
Setup profile revision 21 publishes the AI review's verdict as a check run, and review status reads it
Amends spec 002's setup profile, spec 024's review and spec 006's command surface. github-actions-rust revision 21 publishes every review result (findings, no-findings, skipped:<class>) as a check run on the reviewed head, named `statecraft-ai-review`, whose output carries the evidence record byte for byte. A new read-only verb, `statecraft-cli review status <pr> [--head <sha>] [--repo <owner/name>] [--json]`, reads it under one rule for which review counts, the rule ci-gate already uses in the merge queue, and answers with a closed outcome and spec 006's exit codes. The `## AI review` comment is unchanged. Arming, retrying and waiting are out of scope. Spec 044 moves from revision 21 to revision 22.
Read the full pinned specification ↗Relationships
depends on
depended on by
amends
amended by
Claimed units
- extends
{ "nature": "additive", "spec": "002-environment-lifecycle", "unit": { "kind": "directory", "path": "crates/statecraft-home/" } } - extends
{ "nature": "additive", "spec": "006-command-surface", "unit": { "kind": "directory", "path": "crates/statecraft-cli/" } }
Source headings
- 062: The AI review's verdict as a check run, read by `review status`
- 1. Purpose
- 2. Territory
- 3. Behavior
- 3.1 The verdict check run
- 3.2 Which review counts
- 3.3 The `review status` verb
- 3.4 Outcomes and exit codes
- 3.5 What does not change
- 3.6 The revision
- 4. Out of scope
- 5. Observable negative cases
- 6. Resolved decisions
- Implementation record
- Verification
Frozen anchors
No frozen anchors declared.