PR Review Summary

Status: Drop+Supersede

🪜 Strategic-Fit Decision

Per §9 Strategic-Fit Step-Back:

• Decision: Drop+Supersede
• Rationale: Cycle-1 premise pre-flight fires on a premise-invalid, ticket-prescription-off slice. The patch adds a connection contract which no production caller supplies, misidentifies four client/transport/domain classifications as wire-named states, and lets one global classification relabel another surface's retained reason. Approve+Follow-Up is unavailable because this head has no independently valuable day-after-merge effect and the dormant contract is itself unsafe to consume.
• Disposition: ticket-prescription-off
• Source-coordinate falsifiers: At exact head 1d07bb935f2acbe2ef333d126b5699cf194c11a0, FleetCockpit.mjs:2951-2960 calls deriveSpineBanner without connection; exact-tree search finds production transport and caller controls but every connection: assignment only in the new reducer specs. fleetWireMethods.mjs:110-118 contains only refused of the five advertised values. boundedRead at FleetCockpit.mjs:94-121 rejects at the bound and fences late answers, so the client cannot observe the documented “answers past the read bound” fact.
• Salvage map: Preserve the candidate operator copy and cold/degraded matrix cases as design material. Discard the standalone optional global connection parameter, the all-five-wire-named claim, and this render-only close target. Re-land only with an owner-produced, sanitized, deciding-surface-bound observation plus the production caller witness.
• Successor landing pad: #16834 remains the leaf, with its corrected next-implementation shape recorded at https://github.com/neomjs/neo/issues/16834#issuecomment-5233814786.
• Successor map citation: https://github.com/neomjs/neo/issues/16834#issuecomment-5233814786

Peer-Review Opening: Phoebe, the wait-vs-investigate distinction is real and the copy work is salvageable. The contract boundary beneath it is not yet real, so preserving this patch would make the successor inherit an unowned axis rather than accelerate it.

🧭 Patch-Blind Premise Snapshot

• Inputs Read Before Patch: #16834, parent #16744, sibling #16824, D#16720 body v12/C3, current dev versions of spineBanner.mjs, FleetCockpit.syncSpineBanner, and both Fleet wire-vocabulary twins, plus RunningTheFleetCockpit.md through Knowledge Base synthesis and the shipped throttle-axis precedent.
• Expected Solution Shape: The owner of each roster/activity read should derive a canonical connection observation from an actual request outcome and bind it to that deciding surface; the pure reducer may then map it to operator-action copy. The view must not hardcode future S2/S3 facts detached from the canonical client/transport/wire translation, and tests must isolate both the pure matrix and the real producer/caller seam.
• Patch Verdict: Contradicts the expected shape. The diff adds only an optional reducer argument and direct-injection tests; the exact production caller is byte-unchanged. The global argument also crosses the pre-existing per-surface reason boundary.
• Premise Coherence: Conflicts with verify-before-assert. An honesty banner cannot turn an unowned global label into “safe to wait,” and contract-first cannot mean “spec-only assignment with no production writer.”

🕸️ Context & Graph Linking

• Target Epic / Issue ID: Resolves #16834
• Related Graph Nodes: #16744, #16824, D#16720, PR #16721
• Origin Session ID: 3c9980a6-df65-4909-912e-e8a761c2d70f

🔬 Depth Floor

Challenge OR documented search (per guide §7.1):

Challenge: The new global axis can combine with the cause of a different surface. The exact-head reducer witness grid=live, stream=stale(reason="activity source not wired"), connection=slow returns “The fleet plane is slow — showing last-known data · activity source not wired · safe to wait.” The producer reason says configuration is missing; the connection prefix tells the operator to wait. Both cannot own the same sentence.

Rhetorical-Drift Audit (per guide §7.4):

• PR description inspected against exact code and cited vocabulary
• Anchor & Echo summaries inspected against the production caller and bounded-read semantics
• Linked throttle precedent inspected at its shipped commit
• Discussion C3 authority inspected independently of the PR body

Findings: Fail. “Wire-named” is true only for refused; the four remaining values are absent from the closed wire enum. “Same contract-before-producer shape as the throttle axis” is also false: the throttle render commit consumed an already-produced end-to-end DTO observation. Finally, “the plane answers past the read bound” is not observable through the current Promise.race/generation-fence path.

🧠 Graph Ingestion Notes

• [KB_GAP]: None. The Knowledge Base located the fail-closed cockpit authority; live code supplied the missing SPINE-specific detail.
• [TOOLING_GAP]: The two-query Memory Core prior-art sweep was a clear miss; exact source and the cited Discussion were used instead.
• [RETROSPECTIVE]: A render contract is not contract-first merely because a spec can inject it. The minimum honest slice includes a production writer, ownership-preserving transport, and a real consumer seam.

🎯 Close-Target Audit

• Close-target identified: #16834
• #16834 is an open leaf with enhancement + ai, not epic
• Commit/body magic targets inspected at exact head

Findings: The magic syntax is valid, but the leaf was written around the already-built render-only slice and therefore encodes the invalid prescription. Keep the leaf open and amend its implementation shape; do not close it with this PR.

📑 Contract Completeness Audit

• Originating ticket contains a Contract Ledger matrix
• Implemented diff matches a source-owned contract

Findings: Missing ledger and no source owner for the new consumed connection input. The successor map folds this into the restart shape rather than creating a second iterative action.

🪜 Evidence Audit

• PR body declares L2 achieved / L2 required
• Exact-head unit matrix and full hosted CI are green
• Evidence reaches a production-consumable contract

Findings: The L2 receipt is valid for direct reducer injection, but the close target defines “delivered” around that injection. It does not witness an application state because the only production caller never supplies the fact. Green execution therefore confirms the dormant branch, not the claimed cockpit capability.

📡 MCP-Tool-Description Budget Audit

Findings: N/A — no MCP/OpenAPI surface is touched.

🔗 Cross-Skill Integration Audit

• Checked the canonical wire twins and their response vocabulary
• Checked the production FleetCockpit consumer
• Checked the shipped throttle-axis precedent
• Found a production predecessor that can fire this new pattern

Findings: No predecessor/writer exists. The new architectural primitive is integrated only with its unit spec.

🧪 Test-Evidence & Location Audit

• Execution evidence: all 17 exact-head hosted checks green at 1d07bb935f2acbe2ef333d126b5699cf194c11a0; author reports the 29/29 surface matrix
• Reviewer falsifier: executed the exact extracted reducer against a cross-surface state/reason mismatch; it produced the false “safe to wait” sentence quoted above
• Test location: the pure reducer spec is correctly placed

Findings: Test placement and execution pass; production reach and ownership do not.

📋 Required Actions

To proceed with the terminal disposition:

• Close this PR unmerged and restart from the successor map on #16834. The next coherent slice must land the real client/transport/wire translation owner, bind the observation to the deciding surface, sanitize its reason, pass it through syncSpineBanner, and prove that caller seam alongside the pure rendering matrix. Do not repair this contract-only head in place.

📊 Evaluation Metrics

• [ARCH_ALIGNMENT]: 30 — the pure reducer is correctly located, but the global unowned axis crosses the module's per-surface truth boundary and has no production integration.
• [CONTENT_COMPLETENESS]: 45 — extensive JSDoc and PR prose exist, but three load-bearing claims contradict the exact source and the consumed contract has no ledger.
• [EXECUTION_QUALITY]: 35 — CI and the direct matrix are green, while the exact reducer accepts a false cross-surface composition and the production path cannot exercise any new branch.
• [PRODUCTIVITY]: 25 — it produces candidate copy/tests but delivers none of the operator-visible connection capability named by the leaf.
• [IMPACT]: 60 — if connected later, this banner controls incident guidance across the cockpit; wrong “wait” versus “investigate” copy has meaningful operational blast.
• [COMPLEXITY]: 35 — two files and one pure reducer keep mechanical complexity low; the hidden ownership/translation problem carries most of the conceptual load.
• [EFFORT_PROFILE]: Quick Win — low mechanical scope with salvageable copy and matrix work, but only after the real producer boundary makes it an actual win.

The useful material should survive; the unowned contract should not.

[review-budget-managed]

• outcome: terminal-drop-supersede
• ordinary-limit: 2
• activation-issue: 15257
• activation-pr: 15307
• activated-at: 2026-07-16T20:54:31Z