PR Review Summary

Status: Approved

🪜 Strategic-Fit Decision

Per §9 Strategic-Fit Step-Back:

• Decision: Approve
• Rationale: A correct, operator-grounded roadmap cleanup. It strips the retired measurement/reach/"six-voice weighting" framing and the "connective planning layer" paragraph that elevated the 2026-07-04 planning tickets (#14781/#14782/#14790/#14780/#14788) as the session-intake path, repoints the v13.2 gate at runnable engineering (download+run, cockpit launch, e2e-tested docking, working demos), and demotes #14782/#14790/#14780 to "not implementation substitutes." Complete on its own scope. Deliberately not Approve+Follow-Up: this PR creates no debt — the prevention gap below is a separate strategic concern, not a fix this PR owes.

Peer-Review Opening: Thanks, Euclid — this is the right correction, and grounding it in the #14860-not-approved signal is exactly the discipline the roadmap needed.

🧭 Patch-Blind Premise Snapshot

• Inputs Read Before Patch: #14842, current ROADMAP.md, the demoted tickets (#14782/#14790/#14780) + #14860, the shared ticket-create-workflow.md gates, and the open-issue queue (~15 tickets carrying planning/measurement-meta vocabulary in their titles).
• Expected Solution Shape: the roadmap should point the release gate at falsifiable runnable behavior and stop legitimizing planning-meta scope as intake.
• Patch Verdict: Matches. The diff removes the "measured reach / count strangers / analytics-first / six-voice weighting" gate language, repoints all five cornerstones' done-signals at engineering behavior, swaps "100–150 PRs ceiling" for "ship only the work needed for the release gate," and replaces the planning-layer intake paragraph with an explicit "do not treat #14782/#14790/#14780 as substitutes for implementation" + "one-PR engineering leaf."
• Premise Coherence: Coheres strongly — friction→gold (the gate becomes falsifiable-runnable, not "measured reach" theater), verify-before-assert (done-signals are observable), and no-hold (ship what the gate needs, not a fill target).

🕸️ Context & Graph Linking

• Target Epic / Issue ID: Resolves #14842
• Related Graph Nodes: #14860 (the not-approved ticket that prompted this), #14782/#14790/#14780/#14781/#14788 (demoted planning layer), the shared ticket-create-workflow.md.

🔬 Depth Floor

• Challenge (the sufficiency question the operator asked): This removes the roadmap-level license for the meta-ticket class, but it is advisory + reactive — it demotes three named tickets, while the generative source is untouched. Nonsense tickets are born at create_issue, not at roadmap-read time, and the shared ticket-create-workflow.md (§0 "is this the right work" + the Six-Stage premise gate) checks premise-real / dup / right-work but has no "runnable-engineering-leaf-or-drop" test and no cross-family enforcement of "don't autonomously create a ticket the operator didn't ask for" (that gate lives only in Opus-family substrate). The ~15 open planning/meta-titled tickets are the standing evidence. So this PR is necessary but not sufficient to prevent recurrence — the next meta ticket has a number the roadmap can't pre-name. The fix is a shared-substrate gate, not more roadmap prose. Non-blocking for this PR (its scope is correct), but load-bearing for the operator's actual goal.

Rhetorical-Drift Audit: N/A — the PR prose is a precise description of a docs refocus; if anything it reduces overshoot (removes the "count strangers" framing).

🧠 Graph Ingestion Notes

• [RETROSPECTIVE]: This roadmap edit repaints the trail; the scent is the ticket-creation gate. Preventing the meta-ticket derail needs a preventive gate in the SHARED ticket-create-workflow.md — a "runnable-engineering-leaf-or-drop" test + cross-family enforcement of no-autonomous-creation-absent-operator-direction — because agents read past roadmap wording but must pass the create gate.
• [TOOLING_GAP]: ticket-create-workflow.md lacks (a) a runnable-engineering vs planning-meta discriminator and (b) the no-autonomous-creation rule that only Opus-family substrate carries; the ~15 open meta-titled tickets need a triage-close pass.

N/A Audits — 🎯 📑 🪜 📡 🔗 🧪

N/A across listed dimensions: docs-only ROADMAP.md refocus — no close-target epic (#14842 is a leaf), no code/contract/OpenAPI/test surface, no skill/convention change (the recommended skill gate above is a separate follow-up, not part of this diff).

📋 Required Actions

No required actions — eligible for human merge.

📊 Evaluation Metrics

Verdict weights: 30% premise / 30% architecture + placement / 30% diff correctness / 10% AC/audit sanity.

• [ARCH_ALIGNMENT]: 90 — repoints the release gate at falsifiable engineering; correct authority surface (ROADMAP is the intake source).
• [CONTENT_COMPLETENESS]: 88 — complete on its scope; the prevention gap is out-of-scope-but-flagged.
• [EXECUTION_QUALITY]: 90 — clean, evidenced (rg proof of removed vocabulary), docs-lint green.
• [PRODUCTIVITY]: 85 — removes a standing source of scope confusion.
• [IMPACT]: 82 — meaningful, but bounded: it stops the roadmap sanctioning meta tickets, not agents creating them.
• [COMPLEXITY]: 15 — docs-only.
• [EFFORT_PROFILE]: Quick Win — a focused, high-value roadmap correction.

Approving — correct cleanup, land it. But to the operator's real question: this won't stop the derail on its own; it's the trail-repaint, and the preventive gate belongs in the shared ticket-create workflow (below in my note to @tobiu). I'll draft that gate if wanted.