Context
PR authors already publish their Neo Memory Core origin session in the canonical pull-request body, which gives later peers a direct reasoning trail. The three committed PR-review template assets omit the reviewer's equivalent provenance, so a peer reading a formal review cannot jump directly from that artifact to the reviewer session that produced it.
The operator observed this as a friction-to-gold quick win on 2026-07-29. Memory Core prior-art also surfaced the critical identity boundary: this must be the peer-discoverable Neo Memory Core session UUID, not a harness transcript identifier.
Live latest-open sweep: checked the latest 20 open issues created-descending at 2026-07-29T22:03Z; no equivalent ticket found. The immediately preceding 30-message A2A claim sweep also found no overlapping lane. Semantic KB, Memory Core, exact GitHub search, and resources/content/{issues,discussions} sweeps found prior template-fidelity machinery but no owner for reviewer-session provenance.
The Problem
The provenance contract is asymmetric:
- Pull-request authorship guidance carries
Session <Origin Session ID>.
- The review-response template carries
Origin Session ID.
- Formal review templates do not carry the reviewer's session.
validate_pr_review_body and the GitHub review-body lint enforce template shape, but therefore cannot prevent this provenance field from being omitted or left as a non-queryable placeholder.
Later peers can search semantically, but must reconstruct reviewer intent instead of querying the exact Neo Memory Core session. That is avoidable retrieval friction and weakens the review artifact as graph-ingestion substrate.
The Architectural Reality
Exactly three PR-review template assets exist:
.agents/skills/pr-review/assets/pr-review-template.md.agents/skills/pr-review/assets/pr-review-followup-template.md.agents/skills/pr-review/assets/pr-review-micro-delta-template.md
The full and follow-up shapes are validated by ai/services/github-workflow/PullRequestService.mjs; micro-delta has its own state-vector validation there. The mirrored GitHub enforcement lives in .github/workflows/agent-pr-review-body-lint.yml. Focused fixtures live in test/playwright/unit/ai/services/github-workflow/PullRequestService.spec.mjs.
Progressive-disclosure placement: the three assets are conditional Atlas payloads. .agents/skills/pr-review/SKILL.md is the always-loaded Map and should remain unchanged. Structure-map V-B-A confirms the validator belongs with its existing ai/services/github-workflow siblings; no new .mjs file or service is needed.
The Fix
Add one consistently named Origin Session ID field to each of the three template assets:
- Full review: provenance beside Context & Graph Linking.
- Follow-up: provenance beside the prior/current review anchors.
- Micro-delta: provenance in the State Vector.
The value must be a concrete Neo Memory Core UUID. Extend both existing validation layers to reject a missing field, a template placeholder, or a non-UUID value for each of the three documented formats. Update the existing unit/workflow fixtures and add focused negative coverage.
Contract Ledger
| Target Surface |
Source of Authority |
Proposed Behavior |
Fallback / Edge Case |
Docs |
Evidence |
pr-review-template.md |
This ticket + current pull-request provenance precedent |
Requires Origin Session ID with the reviewer's Neo Memory Core UUID |
Placeholder or non-UUID remains invalid at submission |
Template asset |
Full-body positive + missing/malformed negative validator tests |
pr-review-followup-template.md |
This ticket + existing follow-up skeleton contract |
Carries the current follow-up review session UUID, independent of prior review comment/session anchors |
Missing or malformed value fails preflight/lint |
Template asset |
Follow-up positive + negative validator tests |
pr-review-micro-delta-template.md |
This ticket + existing micro-delta State Vector |
Carries the current micro-delta review session UUID in the State Vector |
Missing or malformed value fails the micro-delta path |
Template asset |
Micro-delta positive + negative validator tests |
PullRequestService.validatePrReviewBody() / managed review guard |
Existing PullRequestService.mjs template-fidelity validator |
Accepts all three documented shapes only with a concrete UUID-shaped origin session |
Returns structured template-validation failure before GitHub dispatch |
JSDoc stays current |
Focused Playwright unit suite |
| Agent PR review-body workflow lint |
Existing .github/workflows/agent-pr-review-body-lint.yml parity layer |
Mirrors the service rule for gate-bearing reviews |
Posts the existing corrective lint comment and fails the check |
Workflow comments stay current |
Workflow-harness unit fixtures |
Decision Record impact
None. This aligns the existing review-provenance and template-fidelity contracts; it does not change an ADR.
Acceptance Criteria
Out of Scope
- Adding a new MCP tool or changing Memory Core APIs.
- Resolving or dereferencing the UUID against Memory Core during GitHub lint.
- Retrofitting historical reviews.
- Creating a fourth review template or expanding the blast-scaled shorthand format.
- Changing PR-author or author-response provenance formats.
Avoided Traps
- A template-only change would be advisory and drift immediately; both existing enforcement layers must move with it.
- Accepting any free-form session token would preserve the harness-ID ambiguity rather than fix retrieval.
- Adding provenance mechanics to the always-loaded
SKILL.md would bloat the Map for a conditional review-composition detail.
Related
Related: #14688, #13547
Origin Session ID: 13ebbc76-8321-4446-8d0b-5f044e145bcd
Retrieval Hint: PR review Origin Session ID reviewer provenance three templates validate_pr_review_body
Context
PR authors already publish their Neo Memory Core origin session in the canonical pull-request body, which gives later peers a direct reasoning trail. The three committed PR-review template assets omit the reviewer's equivalent provenance, so a peer reading a formal review cannot jump directly from that artifact to the reviewer session that produced it.
The operator observed this as a friction-to-gold quick win on 2026-07-29. Memory Core prior-art also surfaced the critical identity boundary: this must be the peer-discoverable Neo Memory Core session UUID, not a harness transcript identifier.
Live latest-open sweep: checked the latest 20 open issues created-descending at 2026-07-29T22:03Z; no equivalent ticket found. The immediately preceding 30-message A2A claim sweep also found no overlapping lane. Semantic KB, Memory Core, exact GitHub search, and
resources/content/{issues,discussions}sweeps found prior template-fidelity machinery but no owner for reviewer-session provenance.The Problem
The provenance contract is asymmetric:
Session <Origin Session ID>.Origin Session ID.validate_pr_review_bodyand the GitHub review-body lint enforce template shape, but therefore cannot prevent this provenance field from being omitted or left as a non-queryable placeholder.Later peers can search semantically, but must reconstruct reviewer intent instead of querying the exact Neo Memory Core session. That is avoidable retrieval friction and weakens the review artifact as graph-ingestion substrate.
The Architectural Reality
Exactly three PR-review template assets exist:
.agents/skills/pr-review/assets/pr-review-template.md.agents/skills/pr-review/assets/pr-review-followup-template.md.agents/skills/pr-review/assets/pr-review-micro-delta-template.mdThe full and follow-up shapes are validated by
ai/services/github-workflow/PullRequestService.mjs; micro-delta has its own state-vector validation there. The mirrored GitHub enforcement lives in.github/workflows/agent-pr-review-body-lint.yml. Focused fixtures live intest/playwright/unit/ai/services/github-workflow/PullRequestService.spec.mjs.Progressive-disclosure placement: the three assets are conditional Atlas payloads.
.agents/skills/pr-review/SKILL.mdis the always-loaded Map and should remain unchanged. Structure-map V-B-A confirms the validator belongs with its existingai/services/github-workflowsiblings; no new.mjsfile or service is needed.The Fix
Add one consistently named
Origin Session IDfield to each of the three template assets:The value must be a concrete Neo Memory Core UUID. Extend both existing validation layers to reject a missing field, a template placeholder, or a non-UUID value for each of the three documented formats. Update the existing unit/workflow fixtures and add focused negative coverage.
Contract Ledger
pr-review-template.mdOrigin Session IDwith the reviewer's Neo Memory Core UUIDpr-review-followup-template.mdpr-review-micro-delta-template.mdPullRequestService.validatePrReviewBody()/ managed review guardPullRequestService.mjstemplate-fidelity validator.github/workflows/agent-pr-review-body-lint.ymlparity layerDecision Record impact
None. This aligns the existing review-provenance and template-fidelity contracts; it does not change an ADR.
Acceptance Criteria
Origin Session IDfield for the reviewer's Neo Memory Core session UUID.validate_pr_review_body.PullRequestService.mjsand.github/workflows/agent-pr-review-body-lint.ymlenforce equivalent behavior.node ai/scripts/lint/lint-skill-manifest.mjs --base origin/devpasses and the always-loadedpr-review/SKILL.mdbyte delta is zero.Out of Scope
Avoided Traps
SKILL.mdwould bloat the Map for a conditional review-composition detail.Related
Related: #14688, #13547
Origin Session ID: 13ebbc76-8321-4446-8d0b-5f044e145bcd
Retrieval Hint:
PR review Origin Session ID reviewer provenance three templates validate_pr_review_body