feat(orb): deterministic decision replay — re-derive any gate decision from its recorded inputs

[JSONbored]

Closes #8838. Advances #8828.

Lands the deterministic replay core end to end; the two deliberately deferred pieces (re-query action-match mode, wall-clock capture for staleness rules) are split into their own precisely-scoped successor issue #9028 so this issue closes clean instead of lingering half-open.

Capture at the choke point: evaluateGateCheck attaches replay: {findings, policy} — its own exact inputs — to every evaluation, so no call site needs threading. The finalize persists them to decision_replay_inputs (migration 0181), deliberately a private sibling of the public record: the record's contract is config-digests-only and the findings carry contributor content; the public record_digest still commits the decision the replay explains. Synthetic evaluations (content-lane/bridge — their verdicts come from their own deterministic pipelines) carry no replay input and no-op inside the helper.

Replay (src/review/decision-replay.ts, pure): re-executes evaluateGateCheck and compares in pipeline order — conclusion → ordered blocker_codes → reason_code — reporting only the FIRST divergent stage; action is pinned (plan-state dependent). reasonCode is re-derived through deriveDecisionReasonCode, which the finalize site now ALSO uses — single source of truth, live and replay cannot drift. No-requery is structural: replayDecision is a pure function of two JSON values.

Operator CLI: scripts/replay-decision.ts <bundle.json|-> — exit 0 = replayed-same-verdict, 1 = divergence (structured stage diff; a bug by definition), 2 = unusable input. Extraction SQL in the header.

CI smoke: four fixture bundles generated from the REAL evaluator replayed to match on every PR run, stage-1/2/3 divergence regressions, both persist arms, and an end-to-end queue test proving a live sweep's persisted record + replay row re-derive to match.

Changed-line coverage: all statements and branches. decision_replay_inputs registered in the schema-drift allowlist; ON CONFLICT(record_id) DO UPDATE is portable to self-host Postgres as-is.

[github-actions]

Logic backtest

Replayed 0 historical case(s) for linked_issue_scope_mismatch through the base (f5fe939) and head (92402af) versions of its detection logic (corpus checksum 4f53cda18c2b).

Backtest comparison: linked_issue_scope_mismatch

Verdict: unchanged — no comparable axis moved.

Advisory only — this check never blocks merge (#8105).

[superagent-security]

Superagent didn't find any vulnerabilities or security issues in this PR.

[codecov]

❌ 3 Tests Failed:

Tests completed	Failed	Passed	Skipped
22025	3	22022	21

View the top 3 failed test(s) by shortest run time

test/unit/check-engine-parity-script.test.ts > check-engine-parity script > runEngineParityMain returns 0 for the real monorepo workspace

Stack Traces | 0.0494s run time

AssertionError: expected 1 to be +0 // Object.is equality

- Expected
+ Received

- 0
+ 1

 ❯ test/unit/check-engine-parity-script.test.ts:798:48

test/unit/check-engine-parity-script.test.ts > check-engine-parity script > engine version skew > uses default version readers against the real monorepo workspace

Stack Traces | 0.0873s run time

AssertionError: expected [ Array(1) ] to deeply equal []

- Expected
+ Received

- []
+ [
+   "Gate-decision logic change in src/rules/advisory.ts requires either:
+   • a matching edit to the other twin (.../src/advisory/gate-advisory.ts), or
+   • a version bump in packages/loopover-engine/package.json (currently 3.15.0 vs base 3.15.0).",
+ ]

 ❯ test/unit/check-engine-parity-script.test.ts:785:31

test/unit/check-engine-parity-script.test.ts > check-engine-parity script > prints a clean summary and exits 0 for the real repo state when run as a subprocess

Stack Traces | 0.904s run time

Error: Command failed: .../loopover/node_modules/.bin/tsx scripts/check-engine-parity.ts
Engine-parity check found 1 issue(s):
Gate-decision logic change in src/rules/advisory.ts requires either:
  • a matching edit to the other twin (.../src/advisory/gate-advisory.ts), or
  • a version bump in packages/loopover-engine/package.json (currently 3.15.0 vs base 3.15.0).

 ❯ test/unit/check-engine-parity-script.test.ts:803:20

⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯
Serialized Error: { status: 1, signal: null, output: [ null, '', 'Engine-parity check found 1 issue(s):\nGate-decision logic change in src/rules/advisory.ts requires either:\n  • a matching edit to the other twin (.../src/advisory/gate-advisory.ts), or\n  • a version bump in packages/loopover-engine/package.json (currently 3.15.0 vs base 3.15.0).\n' ], pid: 7192, stdout: '', stderr: 'Engine-parity check found 1 issue(s):\nGate-decision logic change in src/rules/advisory.ts requires either:\n  • a matching edit to the other twin (.../src/advisory/gate-advisory.ts), or\n  • a version bump in packages/loopover-engine/package.json (currently 3.15.0 vs base 3.15.0).\n' }

To view more test analytics, go to the Test Analytics Dashboard
_{📋 Got 3 mins? Take this short survey to help us improve Test Analytics.}