feat(review): add per-repo review.selftune force-off for the auto-tune cron

[JSONbored]

Summary

• Investigated Add config-as-code coverage for the selftune feature #4104's original "mechanical migration onto resolveConvergedFeature" framing and found it doesn't fit: unlike rag/reputation/grounding, runSelfTune's cron pass has no GITTENSORY_REVIEW_REPOS allowlist to fall back to — its existing repo scoping is isAgentConfigured (whether a repo already opted into the broader acting-autonomy surface), a different consent boundary entirely.
• Added review.selftune: boolean | null as its own top-level FocusManifestReviewConfig field (mirroring impactMap/cultureProfile's plumbing — NOT the features:/CONVERGED_FEATURE_KEYS block, which has no allowlist-fallback shape that fits here).
• Deliberately FORCE-OFF-ONLY — mirroring the safety feature's asymmetric precedent (fix(review): features.safety is per-repo disableable via .gittensory.yml even when the operator enabled it globally #2269) in the opposite direction: an explicit review.selftune: false excludes an otherwise agent-configured repo from the tuning pass. There is no true override — forcing a NON-agent-configured repo INTO tuning would bypass that repo's separate, broader autonomy-consent boundary, which this config key must not touch.
• Wired into selfTuneRepos (src/review/selftune-wire.ts) — the per-repo filter function the cron already uses, with a manifest-load error failing open (repo stays included), matching the existing settings-blip fail-safe.
• Part of Epic: config-as-code should fully control every review feature, not just restrict it #4092 (config-as-code epic). Closes Add config-as-code coverage for the selftune feature #4104.

Scope

• Conventional Commit title.
• Focused change — no unrelated backend/UI/MCP/docs changes.
• Follows CONTRIBUTING.md.
• Linked issue (Add config-as-code coverage for the selftune feature #4104).

Validation

• git diff --check
• npm run typecheck — clean.
• Targeted tests: test/unit/focus-manifest.test.ts, test/unit/config-templates.test.ts, test/unit/signals-coverage.test.ts, test/unit/selftune-wiring.test.ts (3 new tests: force-off regression, unset-stays-default regression, true-override-is-a-no-op), test/unit/selftune-readback.test.ts — all green (652 tests).
• Scoped coverage check — src/review/selftune-wire.ts at 100% line/branch; packages/gittensory-engine/src/focus-manifest.ts's uncovered lines confirmed unrelated to this diff (pre-existing labelingRules/content-parsing gaps).
• npm audit --audit-level=moderate — 0 vulnerabilities.

Safety

• No secrets/wallets/hotkeys/trust-scores exposed.
• N/A — no auth/session/CORS changes.
• N/A — no API/OpenAPI/MCP surface touched.
• N/A — no UI changes.
• This touches the self-improvement/auto-tune safety contract's REPO SCOPING only — the tightening-only shadow-soak/promotion logic itself (computeTuningRecommendations, runAutoApplyRecommendations) is completely untouched.

[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	8d76894	Commit Preview URL Branch Preview URL	Jul 08 2026, 02:10 AM

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 93.65%. Comparing base (e86933f) to head (8d76894).
✅ All tests successful. No failed tests found.

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #4118   +/-   ##
=======================================
  Coverage   93.65%   93.65%           
=======================================
  Files         384      384           
  Lines       35860    35865    +5     
  Branches    13158    13160    +2     
=======================================
+ Hits        33583    33589    +6     
  Misses       1618     1618           
+ Partials      659      658    -1     

Files with missing lines	Coverage Δ
packages/gittensory-engine/src/focus-manifest.ts	99.15% <100.00%> (+<0.01%)	⬆️
src/review/selftune-wire.ts	100.00% <100.00%> (+3.84%)	⬆️

🚀 New features to boost your workflow:

• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[loopover-orb]

Tip

🟩🟩🟩🟩🟩🟩🟩🟩🟩🟩🟩🟩

✅ Gittensory review result - approve/merge recommended

_{Review updated: 2026-07-08 02:19:41 UTC}

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

✅ Suggested Action - Approve/Merge

• safe to merge

Review summary
Adds `review.selftune: boolean | null` as a top-level FocusManifestReviewConfig field (mirroring the impactMap/cultureProfile plumbing pattern) and wires it into `selfTuneRepos` in src/review/selftune-wire.ts as a force-off-only per-repo exclusion from the auto-tune cron. The asymmetric design (no `true` override) is correctly implemented — `manifest?.review.selftune === false` is the only branch that acts, so a `true` value is genuinely a no-op that cannot bypass the `isAgentConfigured` autonomy boundary, and a manifest-load failure correctly fails open via `.catch(() => null)`. All the mechanical plumbing (parse, overlay, JSON serialization, EMPTY_MANIFEST/emptyManifest defaults, field-exhaustiveness test) is updated consistently with the existing convention, and the three new tests exercise the real force-off, unset-default, and true-no-op paths rather than fabricated states.

Nits — 6 non-blocking

• src/review/selftune-wire.ts: `loadRepoFocusManifest` failures are swallowed unconditionally via `.catch(() => null)` — this fails open for real manifest-parse bugs too, not just transient blips, with no logging, so a silent regression here would be invisible in production; consider logging on catch even though the fail-open behavior itself is correct.
• src/review/selftune-wire.ts: the per-repo `loadRepoFocusManifest` await happens sequentially inside the `for` loop for every agent-configured repo, adding one more serial round-trip per repo on top of `resolveRepositorySettings`; fine for a cron job over a bounded repo set today, but worth a comment if the registered-repo count is expected to grow significantly.
• The doc comments in .gittensory.yml.example/gittensory.full.yml/focus-manifest.ts are quite long/repetitive across four locations for a single boolean field — consider trimming to one canonical explanation and linking to it, though this matches existing repo convention for other review.* toggles.
• Consider extracting the `manifest?.review.selftune === false` check into a small named predicate (e.g. `isSelfTuneOptedOut(manifest)`) in selftune-wire.ts for readability, since the surrounding comment already carries most of the semantic weight.
• If `loadRepoFocusManifest` failures are expected to be rare but meaningful, add a debug/warn log on the catch so a systemic manifest-loader issue doesn't silently masquerade as 'every repo passed its selftune check'.
• 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	#4104
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: 51 registered-repo PR(s), 43 merged, 509 issue(s).
Contributor context	✅ Confirmed Gittensor contributor	JSONbored; Gittensor profile; 51 PR(s), 509 issue(s).
Gate result	✅ Passing	No configured blocker found.

Linked issue satisfaction

Partially addressed
The PR delivers the issue's expected outcome (per-repo control of selftune via .gittensory.yml, with docs and tests) but explicitly rejects the prescribed mechanism: it does not add "selftune" to CONVERGED_FEATURE_KEYS/FEATURE_GLOBAL_FLAG or migrate the call site onto resolveConvergedFeature/convergedFeatureActive, instead introducing a separate top-level, force-off-only review.selftune field with

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, JavaScript, Ruby, Go, Kotlin, MDX, Shell
• Official Gittensor activity: 51 PR(s), 509 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

[superagent-security]

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

[loopover-orb]

Gittensory approves — the gate is satisfied and CI is green.