PR Review Summary

Status: Request Changes

Strategic-Fit Decision

Per §9 Strategic-Fit Step-Back:

• Decision: Request Changes
• Rationale: The substrate change is directionally aligned and the checks are green, but the single commit subject violates AGENTS.md §0 Critical Gate 2: every commit subject must end with the ticket id. This is a mechanical merge blocker independent of the PR body's quality.

Peer-Review Opening: I reviewed this as skill-substrate plus commit-gate compliance. The code-vs-data split rule is a valid friction-to-gold conversion; the branch metadata needs one amend before merge.

Context & Graph Linking

• Target Epic / Issue ID: Resolves #11116
• Related Graph Nodes: epic-review Stage 3.5, ticket-create Stage 3, PR #11114 mixed code/data diff, friction-to-gold

Depth Floor

Challenge: The PR body says the implementation follows the ticket exactly, and the diff mostly supports that. The blocker is not the skill prose; it is the commit subject. Exact-head V-B-A:

8efc7ca17720a6d916001f6a633dfefcb8a73889    docs(skills): codify code-vs-data-migration sub-split discipline

That subject does not end with (#11116), so it fails the no-commit-without-ticket invariant.

Rhetorical-Drift Audit:

• PR description: matches the diff and issue prescription.
• Anchor & Echo summaries: N/A; no code/JSDoc surfaces changed.
• [RETROSPECTIVE] tag: N/A.
• Linked anchors: #11116 and PR #11114 support the code-vs-data migration split rationale.

Findings: Request Changes for commit-gate violation.

Graph Ingestion Notes

• [KB_GAP]: N/A.
• [TOOLING_GAP]: N/A.
• [RETROSPECTIVE]: Code-vs-data migration review friction belongs at the epic/sub-ticket shaping tier, not as a per-commit convention, because squash merge collapses commit structure.

Close-Target Audit

• Close-targets identified: Resolves #11116 in PR body.
• #11116 is not epic-labeled.

Findings: Pass.

Contract Completeness Audit

• Source ticket #11116 prescribes epic-review-workflow.md Stage 3.5 plus a sibling ticket-create-workflow.md reminder.
• Diff implements those two surfaces.

Findings: Pass.

Evidence Audit

• PR body contains an Evidence: declaration line.
• AC4 is explicitly deferred to post-merge/runtime validation.
• Static audit results are consistent with exact-head greps: Stage 3.5 appears in the epic-review payload and Code-vs-Data-Migration appears in ticket-create Stage 3.

Findings: Pass.

Cross-Skill Integration Audit

• epic-review owns the Stage 3.5 audit because this fires during epic-shape review.
• ticket-create gets the earlier sibling reminder because the same split should be caught before a mixed-scope epic reaches review.
• No AGENTS_STARTUP.md update needed; no new skill trigger or new skill folder introduced.
• No manifest/symlink update needed; existing skill payloads only.

Findings: Pass.

Test-Execution & Location Audit

• Branch fetched locally: git fetch origin pull/11934/head:codex-review/11934
• Canonical Location: existing skill payloads under .agents/skills/epic-review/references/ and .agents/skills/ticket-create/references/.
• Related verification: git diff --check origin/dev...codex-review/11934 passed.
• Related verification: node ai/scripts/lint/lint-skill-manifest.mjs --base origin/dev passed.
• Related verification: gh pr checks 11934 is green.

Findings: Checks pass; commit subject still blocks.

Required Actions

To proceed with merging, please address the following:

• Amend the branch commit subject so it ends with (#11116), for example: docs(skills): codify code-vs-data-migration sub-split discipline (#11116). The current subject lacks the ticket id and violates AGENTS.md §0 Critical Gate 2.

Evaluation Metrics

• [ARCH_ALIGNMENT]: 95 - 5 points deducted because the substrate tier is right, with only minor concern that the ticket-create reminder is dense inline prose.
• [CONTENT_COMPLETENESS]: 90 - 10 points deducted because the PR body is complete, but commit metadata omits the ticket id.
• [EXECUTION_QUALITY]: 80 - 20 points deducted because CI and local checks pass, but the commit gate is a hard mechanical defect.
• [PRODUCTIVITY]: 85 - 15 points deducted because the ticket's substrate change is delivered but cannot merge until commit metadata is corrected.
• [IMPACT]: 45 - Moderate impact: this prevents future mixed code/data epic-shape review friction across ticket-create and epic-review.
• [COMPLEXITY]: 35 - Low-to-moderate complexity: two skill payload surfaces and cross-skill trigger ordering.
• [EFFORT_PROFILE]: Maintenance - Narrow but durable workflow hygiene.

Once the commit subject is amended, this should be a quick re-review; I did not find a substantive content blocker in the skill prose.

PR Review Follow-Up Summary

Status: Drop+Supersede (formal Request Changes)

Cycle: Cycle 2 corrective re-review

Opening: Prior review focused on commit metadata and treated #11116 as current authority. That was a V-B-A miss: #11116 is an older prescription, while current Map/Atlas substrate changes the strategic-fit decision.

Strategic-Fit Decision

Per pr-review-guide.md §9 Strategic-Fit Step-Back:

• Decision: Drop+Supersede
• Rationale: #11116 was opened on 2026-05-10 and has not been updated since. Current authority now says workflow files are also Maps: hot workflow paths should keep the bare always-relevant minimum inline, move edge-case detail behind one-line trigger pointers, and prefer deletion/compression before adding substrate. This PR follows the old prescription exactly by adding a full Stage 3.5 block plus a dense ticket-create reminder into routine workflow payloads, so green CI only proves the branch is under current mechanical budgets; it does not prove the new prose earns a hot-path slot.

Prior Review Anchor

• PR: #11934
• Target Issue: #11116
• Prior Review Anchor: neo-gpt review submitted 2026-05-24T21:36:00Z
• Author Response Comment ID: N/A
• Latest Head SHA: 8efc7ca17

Delta Scope

This is not an author-code delta; it is a reviewer premise correction after operator calibration.

• Files changed: .agents/skills/epic-review/references/epic-review-workflow.md, .agents/skills/ticket-create/references/ticket-create-workflow.md
• PR body / close-target changes: unchanged; still says "None - implementation follows ticket prescription exactly."
• Branch freshness / merge state: open against dev; CI green at recheck.

Evidence rechecked:

• #11116 live issue: created 2026-05-10, labels enhancement, ai, model-experience, body prescribes this exact Stage 3.5 shape.
• Diff: 2 files, 21 additions. epic-review-workflow.md grows from 18,822 to 21,323 bytes; ticket-create-workflow.md grows from 16,719 to 17,469 bytes.
• git diff --check origin/dev...codex-review/11934 passes.
• gh pr checks 11934 --watch=false passes all current checks.
• Current authorities: create-skill/references/skill-authoring-guide.md "Map vs World Atlas" and recursive workflow-map rule; ADR 0008 §2.2/§2.7/§5.5; ADR 0007 §5.4 additive-drift reflex.

Previous Required Actions Audit

• Still open but no longer primary: amend commit subject to end with (#11116). This remains a hard gate if the branch is preserved, but it is not the right strategic blocker anymore.
• Superseded premise: prior review said the skill prose had no substantive content blocker. I withdraw that finding. The missing check was source-ticket currency against current Map/Atlas and workflow-bloat authority.

Delta Depth Floor

Delta challenge: The PR treats "matches #11116 exactly" as sufficient. For stale workflow-substrate tickets, that is not enough. Current doctrine requires classifying the lesson before writing, then choosing keep, move, compress-to-trigger, rewrite, or retire; the default for an edge-case audit inside a workflow Map is not another inline section.

Conditional Audit Delta

Source-of-Authority Audit

The operator correction prompted the re-review, but the blocking rationale is not "operator preference." The blocking rationale is repo-local authority:

• skill-authoring-guide.md: workflow files are also Maps; edge-case sections extract behind one-line triggers.
• ADR 0008: Map/Atlas applies recursively to workflow files; lesson promotion should compress runtime behavior into Bias/Rule/Rationale/Trigger and move incident history/provenance behind Atlas pointers.
• ADR 0007: when a substrate-correction ticket exists to reduce friction, adding another audit/checklist/template is the wrong default if deletion or compression preserves behavior.

Findings: Request Changes. #11116 is input evidence, not sufficient authority for this 2026-05-24/25 PR shape.

Cross-Skill Integration Audit

The proposed rule may still be valuable. The problem is placement and density:

• The new epic-review Stage 3.5 includes trigger text, audit instructions, sequencing rule, findings behavior, empirical anchor, operator provenance, and out-of-scope rationale inline in the hot workflow.
• The ticket-create reminder is a single long inline rule carrying trigger, rule, rationale, cross-skill pointer, and incident provenance.
• Neither surface applies the current Lesson Promotion atom or slot-disposition analysis before adding runtime text.

Findings: Progressive Disclosure / Map-vs-Atlas blocker.

Test-Execution & Location Audit

• Changed surface class: skill documentation / workflow substrate.
• Location check: existing files under .agents/skills/**/references/; no new test location involved.
• Related verification run: git diff --check origin/dev...codex-review/11934 passed; gh pr checks 11934 --watch=false passed all checks.
• Findings: Mechanical checks pass. Strategic substrate fit still blocks.

Metrics Delta

Metrics replace the prior review because the prior source-of-authority premise was wrong.

• [ARCH_ALIGNMENT]: 45 - 50 points deducted from the prior 95 because the PR now conflicts with current recursive Map/Atlas authority and additive-drift discipline.
• [CONTENT_COMPLETENESS]: 55 - 35 points deducted from the prior 90 because the PR body claims no deltas from #11116 but does not document source-ticket currency, slot disposition, or current Map/Atlas rationale.
• [EXECUTION_QUALITY]: 75 - 5 points deducted from the prior 80 only for the already-known commit-subject gate; the new blocker is strategic fit, not syntax or CI.
• [PRODUCTIVITY]: 30 - 55 points deducted from the prior 85 because exact delivery of a stale prescription moves the substrate in the wrong direction.
• [IMPACT]: 50 - Moderate: code-vs-data split discipline can be valuable, but hot workflow-bloat has recurring swarm-wide cost.
• [COMPLEXITY]: 45 - Moderate: the diff is small, but placement crosses two skill workflows and current substrate-compaction doctrine.
• [EFFORT_PROFILE]: Maintenance - Narrow workflow-substrate hygiene with high review-surface sensitivity.

Required Actions

To proceed with merging, please address the following:

• Do not merge this PR in its current "follows #11116 exactly" shape. Run a current-authority refresh and either close/drop this branch in favor of a superseding ticket/PR, or rewrite it to the superseding shape: a minimal one-line trigger pointer in the workflow Map only if the rule still earns a hot-path slot, with heavy audit/provenance moved to an Atlas payload; apply ADR 0007 slot disposition and ADR 0008 Lesson Promotion atom in the PR body. If you preserve this branch after that reshape, the commit subject still must be amended to end with (#11116) or the superseding ticket id.

Graph Ingestion Notes

• [KB_GAP]: The prior review incorrectly accepted an old source ticket as authority without checking newer Map/Atlas substrate. This is the same class as stale-ticket/currency-risk in pr-review-guide.md §9.0.
• [TOOLING_GAP]: Current CI catches byte-budget and section-trigger extremes, but not small additive workflow-map drift. Human/agent review still has to apply ADR 0007/0008 slot-disposition judgment.
• [RETROSPECTIVE]: "Matches ticket exactly" is not a merge argument when the ticket is older than newer workflow-bloat doctrine. For skill substrate, PR review must validate current authority before treating ticket ACs as sufficient.

A2A Hand-Off

After posting this follow-up review, I will send the review URL / ID via A2A so the author can fetch the delta directly.