fix(github): retry label and assignee writes on stale installation tokens

[bohdansolovie]

Summary

• Wraps ensurePullRequestLabel, removePullRequestLabel, and ensurePullRequestAssignee in withInstallationTokenRetry, matching postPullRequestComment, closePullRequest, and the other GitHub-write helpers.
• Stale cached installation tokens that produce a 401 now get one eviction + remint retry instead of failing the label/assignee mutation permanently while leaving later write helpers to recover alone.
• Updates the label-only public-surface queue regression for permission-scope 403 retry (2 writers × 2 attempts).
• Adds regression coverage for both label helpers and the assignee helper (#6191).

Closes #6191

Scope

• Code changes stay inside the declared surfaces (src/github/labels.ts, src/github/assignees.ts, matching unit tests)
• No public API / export surface change beyond fixing helper behavior
• Reuses the existing withInstallationTokenRetry helper — no new retry semantics invented here

Validation

• Matching unit tests updated (github-labels.test.ts, github-assignees.test.ts, queue-4.test.ts)
• Focused regression coverage for the exact failure mode called out in #6191
• CI on this PR (will mark ready after green)

Safety

• Diff is limited to GitHub label/assignee write retries (+ one queue expectation)
• No scheme / economics / protocol changes
• Existing non-401 failure paths (missing privileges, agents, rate limits) still return { applied: false } / throw without inventing success

Made with Cursor

[loopover-orb]

Warning

⏸️ LoopOver review result - manual review recommended

_{Review updated: 2026-07-16 01:06:14 UTC}

5 files · 1 AI reviewer · no blockers · CI green · unstable

⏸️ Suggested Action - Manual Review

Review summary
This PR extends the existing withInstallationTokenRetry wrapper (already used by postPullRequestComment, closePullRequest, and createOrUpdateNamedCheckRun) to ensurePullRequestLabel, removePullRequestLabel, and ensurePullRequestAssignee, closing the gap called out in #6191 where these three helpers minted a token once and failed permanently on a stale-cache 401. The refactor correctly preserves existing dedup/short-circuit logic and the specific error-swallowing branches (422 already_exists, 404 not-on-PR, 403 assigning-agents-not-supported) inside the retried closure so those still return early without triggering an extra remint. The queue-4.test.ts update showing labels going from 2 to 4 confirms that non-401 write failures (a 403 permission-scope error) also trigger a remint+retry under the shared helper, which is pre-existing behavior of withInstallationTokenRetry now correctly inherited by these call sites rather than a new bug.

Nits — 5 non-blocking

• src/github/labels.ts and assignees.ts: wrapping these paths in withInstallationTokenRetry means a genuine (non-stale-token) permission-scope 403 now costs 2x the GitHub API calls and mints a second token before ultimately failing — worth a one-line note in the PR description since it changes rate-limit budget consumption for the permanently-failing case, even though the retry semantics themselves are inherited unchanged from the shared helper.
• test/unit/queue-4.test.ts:6135: the updated comment explains the 2x label-write count but doesn't assert on token-mint count directly in this test, so a future regression in withInstallationTokenRetry's retry trigger wouldn't be caught here specifically.
• test/unit/github-assignees.test.ts: no regression test for the case where the retried token also comes back 401 (permanent failure) to confirm the helper still throws/propagates rather than retrying indefinitely — likely covered by shared withInstallationTokenRetry tests, but worth a pointer comment.
• Consider adding an assertion on tokenMints alongside the labels:4 count in queue-4.test.ts to lock in the exact remint behavior, not just the resulting write count.
• A short note in the PR description on the rate-limit/token-mint cost of retrying legitimate (non-stale) 403 failures would help reviewers reason about worst-case GitHub API usage under sustained permission-scope failures.

Flagged checks (non-blocking)

• Contributor trust — Contributor flagged for review

Decision drivers

• ✅ Code review — No blockers (1 reviewer)
• ✅ Gate result — Passing (No configured blocker found.)

Context & advisory signals — never blocks the verdict

Signal	Result	Evidence
Linked issue	✅ Linked	#6191
Related work	✅ No active overlap found	No same-issue or scoped active PR overlap found.
Change scope	❌ 8/20	High 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: 79 registered-repo PR(s), 28 merged, 6 issue(s).
Contributor context	✅ Confirmed Gittensor contributor	bohdansolovie; Gittensor profile; 79 PR(s), 6 issue(s).
Improvement	✅ Minor	risk: clean · value: minor

Review context

• Author: bohdansolovie
• Role context: outside_contributor
• Public audience mode: oss maintainer
• Lane context: Repository is configured for direct PR review.
• Public profile languages: not available
• Official Gittensor activity: 79 PR(s), 6 issue(s).
• PR-specific overlap: none found.

Contributor next steps

• Start here: Add a concise scope and risk note.
• Then work through the remaining 1 step in the Signals table above.

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.

🧪 Chat with LoopOver

Ask LoopOver a question about this PR directly in a comment — grounded only in the same cached, public-safe facts shown above, never a new claim.

• @loopover ask <question> answers contribution-quality Q&A with source citations and freshness.
• @loopover chat <question> answers in natural prose from cached decision-pack facts via local inference (maintainer/collaborator; read-only).
• A plain-language @loopover mention with a real question is routed to the closest matching read-only command automatically — no exact syntax required.

Full command reference: https://gittensory.aethereal.dev/docs/loopover-commands

🧪 Experimental — new and may change.

_{🟩 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 LoopOver, a quiet PR intelligence layer for OSS maintainers.

• Re-run LoopOver review

[superagent-security]

🚨 Contributor flagged. Click here for more info: Superagent Dashboard

[superagent-security]

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

[bohdansolovie]

@Loopover CI is green on this PR (validate + all test shards + validate-tests-merge). Patch coverage is 100%. Please re-run review with the updated CI status.

[loopover-orb]

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