feat(miner-governor): self-review adapter wiring diffs through predicted-gate + slop (#2334)

[JSONbored]

Summary

• Builds the adapter that takes an attempt's live worktree diff state and turns it into the SAME inputs buildPredictedGateVerdict and the slop-signal pass expect, so the iterate-loop's self-review call (maintainer: local create->score->self-review iterate-loop orchestrator (the control-flow core) #2333, not yet built) is genuinely byte-identical to what the live maintainer gate would compute post-submission — not an approximation.
• changedPaths is always threaded through explicitly, never omitted, so path-dependent predicted-gate checks are genuinely evaluated. An omitted-paths call silently under-predicts per predicted-gate.ts's own PREDICTED_GATE_NOTE_NO_PATHS disclaimer — exactly the false-confidence bug this issue's deliverables call out by name.
• Cross-boundary problem solved with existing precedent, not a new pattern: src/signals/slop.ts has not been extracted into packages/gittensory-engine (it depends on several sibling src/signals/* modules that are also unextracted, and porting all of them is out of this issue's scope). This mirrors the established RewardRiskEngineDeps injection pattern (reward-risk.ts, feat(miner-foundation): extract reward-risk scoring into gittensory-engine #2281, itself solving the identical problem for a different signal): the slop assessment is an injected dependency (SelfReviewAdapterDeps.runSlopAssessment), with SelfReviewSlopInput/SelfReviewSlopAssessment as a hand-kept structural mirror of SlopAssessmentInput/SlopAssessment — same discipline types/predicted-gate-types.ts's own header comment already documents ("Local mirrors from src/... Keep in sync by hand — the engine package cannot import across into src/."). The engine package keeps zero import dependency on the private src/ tree; the real binding happens in whichever src-side shim eventually wires a live iterate-loop.
• passesPredictedGate is the hard requirement the deliverables explicitly call for: true only when predictedGateVerdict.conclusion === "success" (exported as SELF_REVIEW_PASSING_CONCLUSION) — every other conclusion (failure/action_required/neutral/skipped) reads false. This is defense-in-depth: the sibling orchestration-loop issue (maintainer: local create->score->self-review iterate-loop orchestrator (the control-flow core) #2333) is expected to independently enforce the same requirement.
• Tests use real predicted-gate fixtures (mirroring the exact fixture style already established in test/unit/predicted-gate.test.ts): a genuinely passing synthetic diff and a genuinely blocked one (duplicate-PR collision), and assert byte-identical deep equality against calling buildPredictedGateVerdict directly with the same translated inputs — proving the adapter is a faithful pass-through, not an approximation.
• Barrel-exports RepositoryRecord/IssueRecord/PullRequestRecord/BountyRecord/IssueQualityReport/AdvisoryFinding from types/predicted-gate-types.ts — previously internal-only to predicted-gate.ts's own call sites, now needed because SelfReviewContext/SelfReviewSlopAssessment reference them in their own public signature and must be nameable by consumers (verified: IssueQualityReport is confirmed a pure re-export from the same source in both places it was previously imported from, so this isn't introducing a second parallel type).

Scope

• Conventional Commit title.
• Focused: one new engine module + its test + the barrel-export additions its own public types require.
• Follows CONTRIBUTING.md.
• Owner PR — linked issue eligibility gate does not apply. Closes maintainer: self-review adapter wiring iterate-loop diffs through predicted-gate + slop scoring #2334.
• No dependency on the other open Miner Wave 3 PRs — branches directly off main.

Validation

• git diff --check
• npm run typecheck — clean.
• npm run build + npm run test in packages/gittensory-engine — 380/380 passing (9 new self-review-adapter tests: barrel smoke test, input/path/slop-input mapping, a passing real-fixture scenario, a blocked real-fixture scenario each asserted byte-identical against a direct buildPredictedGateVerdict call, an explicit both-branches assertion of the hard pass/fail requirement, a changedPaths threading proof, and a slop-injection-seam proof).
• npm run test:coverage, test:workers, build:mcp, test:mcp-pack, ui:*, npm audit — not re-run locally; this PR touches only packages/gittensory-engine/**, and per CONTRIBUTING.md only src/** is Codecov-measured. CI runs them authoritatively.

Safety

• No secrets, wallet/hotkey/trust-score/reward data anywhere.
• DETECTOR/COMPOSER only — no IO, no submission-triggering, no behavior change for any existing consumer. runSelfReview is not called from any real driving loop yet (that's maintainer: local create->score->self-review iterate-loop orchestrator (the control-flow core) #2333/maintainer: wire the submission-gate trigger into the harness driving loop #2337, later work).
• The "never treat a non-pass as passing" hard requirement is asserted with a dedicated test exercising both branches of the equality check, not just incidentally covered by the pass/blocked fixture tests.
• No UI changes. No docs/schema changes needed (no new config surface).

Notes

• Sequencing: maintainer: iterate-loop stop/abandon/handoff policy + Phase 4 handoff contract #2335 (iterate-loop policy — the pure decideNextAction state machine consuming this adapter's verdict shape) lands next, then maintainer: local create->score->self-review iterate-loop orchestrator (the control-flow core) #2333 (the iterate-loop orchestrator itself, composing the CodingAgentDriver + this adapter + the policy + the attempt-log primitive — all four already exist or are landing in this batch).
• Design call worth flagging for review: I chose to barrel-export the six previously-internal types (RepositoryRecord etc.) rather than keep SelfReviewContext referencing unnameable types. The alternative would have been redefining yet another local mirror inside self-review-adapter.ts itself (a third parallel copy, on top of types/predicted-gate-types.ts's existing mirror of the src-side originals) — barrel-exporting the existing mirror seemed clearly better, but flagging in case there's a reason those six were deliberately kept internal-only that I'm not seeing.

[superagent-security]

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

[cloudflare-workers-and-pages]

Deploying with    Cloudflare Workers

The latest updates on your project. Learn more about integrating Git with Workers.

Status	Name	Latest Commit	Preview URL	Updated (UTC)
✅ Deployment successful! View logs	gittensory-ui	64d49f0	Commit Preview URL Branch Preview URL	Jul 11 2026, 09:39 AM

[loopover-orb]

Tip

🟩🟩🟩🟩🟩🟩🟩🟩🟩🟩🟩🟩

✅ Gittensory review result - approve/merge recommended

_{Review updated: 2026-07-11 09:41:26 UTC}

3 files · 1 AI reviewer · no blockers · readiness 100/100 · CI green · clean

✅ Suggested Action - Approve/Merge

• safe to merge

Review summary
The AI review returned non-blocking notes for this change but did not include a separate narrative summary. Review the nits below before deciding this PR.

Nits — 5 non-blocking

• `SelfReviewSlopInput.tests` (self-review-adapter.ts:69) is declared but never populated by `buildSelfReviewSlopInput` (self-review-adapter.ts:122-131) — either wire it from `diffState` or drop the field so the mirror doesn't imply a mapping that doesn't exist.
• The hand-kept structural mirrors (`SelfReviewSlopInput`/`SelfReviewSlopAssessment`) have no compile-time or runtime check tying them back to the real `SlopAssessmentInput`/`SlopAssessment` in `src/signals/slop.ts` — a future field rename there will silently desync until the `src`-side shim breaks; worth a one-line note on where that shim's own type-check should live.
• Test file imports from `../dist/index.js` rather than source — confirm this matches the existing convention for other `packages/gittensory-engine/test/*.test.ts` files so the build step is a required precondition for `validate-tests` and not accidentally skipped.
• `buildSelfReviewPredictedGateInput` doesn't thread `diffState.issueDiscoveryLane` into anything other than the slop input — confirm `buildPredictedGateVerdict` genuinely has no path-independent use for it, otherwise it's silently dropped for the gate call.
• PR author also opened the linked issue — Link an issue that was opened by a different contributor, or provide a rationale for why this self-authored issue represents genuine discovery work.

Signal	Result	Evidence
Code review	✅ No blockers	1 reviewer
Linked issue	✅ Linked	#2334
Related work	✅ No active overlap found	No same-issue or scoped active PR overlap found.
Change scope	✅ 20/20	Low review scope from cached public metadata (1 linked issue).
Validation posture	✅ 25/25	PR body includes validation/test evidence.
Contributor workload	✅ 10/10	Author activity: 47 registered-repo PR(s), 39 merged, 428 issue(s).
Contributor context	✅ Confirmed Gittensor contributor	JSONbored; Gittensor profile; 47 PR(s), 428 issue(s).
Gate result	✅ Passing	No configured blocker found.
Improvement	✅ Minor	risk: clean · value: minor — Code changes are accompanied by test evidence.

Review context

• Author: JSONbored
• Role context: owner (maintainer lane)
• Public audience mode: oss maintainer
• Lane context: Repository is configured for direct PR review.
• Public profile languages: Python, TypeScript, Ruby, Go, JavaScript, MDX, Shell, Solidity
• Official Gittensor activity: 47 PR(s), 428 issue(s).
• PR-specific overlap: none found.

Contributor next steps

• Treat this as maintainer-lane context rather than normal contributor-lane activity.

Signal definitions

• Related work = same linked issue, overlapping active PRs, or title/path similarity.
• Change scope = cached public metadata such as size labels, draft state, and review-burden hints.
• Validation posture = whether the PR provides enough public validation/test evidence for maintainer review.
• Contributor workload = public contributor activity and cleanup pressure, not a repo-wide quality failure.
• Contributor context = public GitHub/Gittensor identity context; non-Gittensor status is not a blocker.

_{🟩 Safe / merged · 🟦 Advisory · 🟨 Held for review · 🟥 Blocked / closed}

---

💰 Earn for open-source contributions like this. Gittensor lets GitHub contributors earn for the work they already do — register to start earning →.

Checked by Gittensory, a quiet PR intelligence layer for OSS maintainers.

• Re-run Gittensory review