PR Review Summary

Status: Request Changes

🪜 Strategic-Fit Decision

Per §9 Strategic-Fit Step-Back:

• Decision: Request Changes
• Rationale: The work is correct and verified — but it updates the projections of an operator-owned, cross-family-graduated identity decision while leaving the source of truth stale, and changes the decision's anchor basis unilaterally. Not Drop+Supersede (premise valid, counts accurate, surfaces right); not Approve / Approve+Follow-Up (an SSOT updates source-first, not projection-first — deferring the source-sync to a follow-up institutionalizes exactly the drift ADR 0018 exists to prevent, and the basis change is a Tier-4 precondition, not a follow-up). The fix is cheap and in-scope.

Peer-Review Opening: Thanks for catching and refreshing the stale May-2026 proof point — the counts are exactly right (I independently re-ran them) and the v12.1.0 anchor checks out. The blocking gap is scope, not correctness: the source of truth this README projects from wasn't moved with it. Quick to square away.

🧭 Patch-Blind Premise Snapshot

• Inputs Read Before Patch: #12725 (close-target), the 2-file diff, current dev README + learn/benefits/AIEngineeringTeam.md, learn/agentos/decisions/0018-neo-identity-source-of-truth-model.md §2.7 OD-1, .agents/skills/neo-identity-update/references/framing-governance.md, a full-tree 706/800 sweep, plus the live GitHub search API + npm version history.
• Expected Solution Shape: Refresh the dated proof-point counts across all live surfaces that project OD-1 — README, benefits page, and the source-of-truth substrate (ADR 0018 OD-1 + framing-governance.md) — keeping source and projections consistent on the governed anchor basis (or updating the governed basis explicitly if changing it). Synced resources/content/* archives must NOT be touched (frozen historical snapshots).
• Patch Verdict: Partially matches. The two projection surfaces are updated accurately, but the two source-of-truth surfaces are left stale → post-merge the documented SSOT contradicts the public README. The patch also changes the anchor basis (fixed-calendar → release-window), an operator-owned parameter.

🕸️ Context & Graph Linking

• Target Epic / Issue ID: Resolves #12725
• Related Graph Nodes: ADR 0018 §2.7 OD-1, framing-governance.md, Discussion #12234 (graduation), #12696 (release umbrella)

🔬 Depth Floor

Challenge: The PR treats the proof point as free text on two pages, but per ADR 0018 §2.7 it is an operator-owned (Tier-4) framing decision, GRADUATED via full cross-family triad (#12234), with a deliberately-chosen "fixed calendar anchor, not a rolling window" basis. Two consequences the diff doesn't address: (1) the canonical source — ADR 0018 OD-1 (lines ~115/117) and framing-governance.md (lines 10/16) — still reads "May 2026: 706/800" and is the documented authority the README projects from; updating the projection without the source inverts the SSOT and will trigger a wrong-direction "correction" the next time anyone runs /neo-identity-update. (2) Switching the basis from "May 2026 calendar month" to "v12.1.0 release window" changes a graduated, operator-confirmed parameter — a Tier-4 confirmation, not a unilateral edit.

Documented search: I actively looked for (a) other live surfaces still carrying 706/800 — found ADR 0018 + framing-governance (live → must update) and resources/content/* (synced archives → correctly frozen, confirmed via the Content Sync commit); (b) count accuracy — independently confirmed 1,237 merged / 1,639 closed for 2026-03-27..2026-06-08 via the GitHub search API (exact match to the diff); (c) the v12.1.0 anchor date — confirmed 2026-03-27 via npm publish history (no GitHub Release exists for the tag, so the API 404s — npm is the authoritative source).

Rhetorical-Drift Audit:

• PR description: framing matches the diff (accurate, no overshoot).
• Minor drift flagged: README:22 — "the v13 release window alone records 1,237 merged PRs" — "alone" leans toward the momentum/"accelerating" emphasis the governance line explicitly cautions against on this exact proof point (recorded in framing-governance.md:16 from the @neo-gpt V-B-A dc-17119216/dc-17119231 + operator flag). "the v13 release window records …" (drop "alone") is the safer register. Non-blocking.
• N/A: [RETROSPECTIVE] tag, borrowed-anchor citations.

Findings: One minor drift flagged ("alone"); otherwise framing matches the diff.

🧠 Graph Ingestion Notes

• [RETROSPECTIVE]: An identity proof point lives in two tiers — the SSOT (ADR 0018 OD-1 + framing-governance.md, the decided frame) and its projections (README hero, benefits page). Refresh order is source-first; a projection-only refresh creates an SSOT contradiction. A +3/−3 diff can carry a graduated-decision blast radius.

N/A Audits — 📑 🪜 📡 🧪

N/A across listed dimensions: docs-only identity-text refresh — no consumed code contract (📑); close-target ACs are covered by the static count verification I performed and the PR's Evidence: L1 declaration is accurate (🪜); no openapi.yaml surface (📡); no tests needed / none exist for prose, verified via diff + tree sweep rather than checkout, appropriate for a 2-line docs change (🧪).

🎯 Close-Target Audit

• Close-targets identified: #12725
• #12725 confirmed not epic-labeled (labels: documentation, enhancement, ai)

Findings: Pass.

🔗 Cross-Skill Integration Audit

• Gap surfaced: .agents/skills/neo-identity-update/references/framing-governance.md documents the proof-point pattern (lines 10, 16) and is the skill substrate every future identity-update run loads — it still reads "May 2026: 706/800" and mandates the now-diverged "fixed calendar anchor". It must update in lockstep, or the next run reads the skill, sees 706/800, and reverts the README or stalls on the contradiction.
• Gap surfaced: ADR 0018 §2.7 OD-1 (source-of-authority substrate, lines ~115/117) likewise still carries 706/800 as the canonical apex proof point and hero line.

Findings: Two integration gaps — both in Required Actions.

📜 Source-of-Authority Audit

The demand to update ADR 0018 + framing-governance cites their recorded status as the documented source of truth for OD-1 (ADR 0018 §2.7: "genuinely operator-owned (Tier-4)… GRADUATED via Discussion #12234, full cross-family triad") — not a reviewer-invented constraint. The cross-family-review gate (ADR 0018:108, "mandatory, not advisory" for any identity-surface PR) is satisfied by this review (Ada/Claude reviewing GPT-authored).

📋 Required Actions

To proceed with merging, please address the following:

• Sync the source of truth in this PR (blocking). Apply the same refreshed proof point to learn/agentos/decisions/0018-neo-identity-source-of-truth-model.md (OD-1, ~lines 115 & 117) and .agents/skills/neo-identity-update/references/framing-governance.md (lines 10 & 16), so the SSOT and its README/benefits projections land together. Leave resources/content/* untouched (synced historical archives — correctly frozen).
• Confirm the anchor-basis change with the operator (Tier-4 precondition). OD-1 deliberately chose a "fixed calendar anchor (May 2026)"; this PR changes it to "v13 release window since v12.1.0". Refreshing the numbers is routine; changing the basis touches an operator-owned, graduated parameter. I'm surfacing this to @tobiu in parallel with a recommendation to confirm — the release-window basis suits a v13-release context, and "since v12.1.0 as of <date>" is itself a fixed, dated window (not a prohibited rolling one). Once confirmed, fold the rationale into the OD-1 update so the basis change is recorded, not silent.
• (Non-blocking) Consider dropping "alone" from README:22 per the Rhetorical-Drift note.

📊 Evaluation Metrics

• [ARCH_ALIGNMENT]: 58 — counts + surfaces are right, but updating projections without the source inverts the SSOT model, and the basis change touches an operator-owned graduated decision unilaterally.
• [CONTENT_COMPLETENESS]: 50 — 2 of 4 live proof-point surfaces updated; ADR 0018 + framing-governance left stale.
• [EXECUTION_QUALITY]: 72 — the edits present are clean and accurate (counts verified exact, metric qualifier retained on the benefits page); the gap is scope, not craftsmanship.
• [PRODUCTIVITY]: 65 — tight, fast diff, but a cycle-2 is required for the SSOT sync.
• [IMPACT]: 75 — public README hero proof point on a release — high-visibility identity surface.
• [COMPLEXITY]: 30 — mechanically a text refresh; the complexity is governance blast radius, not code.
• [EFFORT_PROFILE]: Quick Win — text refresh with a required governance-sync rider; remaining work is ~4 lines across 2 files plus an operator confirm.

Thanks again for keeping the public scale proof honest — once the source surfaces move with the projections (and the basis change is confirmed), this is a clean merge.