fix(review): meter visual vision BYOK calls

[JSONbored]

Motivation

• The visual-vision path (runVisualVisionForAdvisory) previously decrypted a repo BYOK key and called the external provider without checking the shared per-repo daily BYOK cap, recording usage, or caching, allowing repeated contributor actions to cause maintainer-paid provider calls.
• This change closes that financial/resource-exhaustion hole by reusing the existing BYOK quota/accounting mechanisms used by other AI features.

Description

• Enforce the per-repo BYOK daily limit before fetching screenshots or calling the provider by invoking countByokAiEventsForRepoSince and using AI_BYOK_DAILY_REPO_LIMIT/DEFAULT_BYOK_DAILY_REPO_LIMIT with clampNumber in runVisualVisionForAdvisory (src/queue/processors.ts).
• Record every visual-vision BYOK attempt/result via a new helper recordVisualVisionUsage that calls recordAiUsageEvent with feature: "visual_vision" and model: byok:; provider usage/token/cost fields are forwarded when available so calls count toward the shared BYOK counter (src/queue/processors.ts`).
• Record usage on both successful returns and no-output/provider-failure outcomes so later passes are counted toward the cap (callAiProvider usage is now recorded where applicable).
• Add regression unit tests covering a zero AI_BYOK_DAILY_REPO_LIMIT (ensures no screenshot/provider calls) and verifying successful visual BYOK calls increment the repo/day counter (test/unit/visual-vision-wiring.test.ts).
• Minor typecheck-safe stringification of provider failure info and import adjustments to bring needed helpers into scope.

Testing

• Ran the focused unit suite npx vitest run test/unit/visual-vision-wiring.test.ts, and the new/modified tests passed.
• Ran type checking with npm run typecheck and it passed with no errors.
• Attempted the full coverage run with npm run test:coverage, but the run exposed unrelated long-running/hang behavior in the existing test/unit/queue.test.ts shard and was interrupted (investigation of the unrelated suite noise is out of scope for this patch).
• npm audit --audit-level=moderate could not complete in this environment due to the registry returning 403 Forbidden.

---

Codex Task

[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	cad1ab7	Commit Preview URL Branch Preview URL	Jul 10 2026, 08:26 AM

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 94.11%. Comparing base (efcd25d) to head (00b52c5).
⚠️ Report is 17 commits behind head on main.
✅ All tests successful. No failed tests found.

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #4363   +/-   ##
=======================================
  Coverage   94.11%   94.11%           
=======================================
  Files         432      432           
  Lines       38348    38368   +20     
  Branches    13979    13988    +9     
=======================================
+ Hits        36091    36111   +20     
  Misses       1600     1600           
  Partials      657      657           

Files with missing lines	Coverage Δ
src/db/repositories.ts	96.61% <100.00%> (+<0.01%)	⬆️
src/queue/processors.ts	95.41% <100.00%> (+0.02%)	⬆️

🚀 New features to boost your workflow:

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

[gittensory-orb]

Warning

🟨🟨🟨🟨🟨🟨🟨🟨🟨🟨🟨🟨

⏸️ Gittensory review result - manual review recommended

_{Review updated: 2026-07-10 08:54:45 UTC}

2 files · 1 AI reviewer · 3 blockers · readiness 93/100 · CI green · unstable

⏸️ Suggested Action - Manual Review

• No linked issue detected — If this PR is intended to solve an issue, link it explicitly in the PR body.
• Maintainer requires a linked issue — Link the relevant issue (for example Closes #123) before opening the PR.
• Possible leaked secret in the diff (generic_secret_assignment) — Remove the secret from the diff, rotate the exposed credential, then re-run the gate.

Review summary
This PR closes a real gap: runVisualVisionForAdvisory previously called the BYOK vision provider without checking the shared per-repo daily BYOK cap or recording usage, so it now calls countByokAiEventsForRepoSince against AI_BYOK_DAILY_REPO_LIMIT/DEFAULT_BYOK_DAILY_REPO_LIMIT before fetching screenshots, and records every BYOK attempt (quota-exceeded, no-output/failure, and success) via a new recordVisualVisionUsage helper feeding recordAiUsageEvent. The self-host vision path is correctly left unmetered since it consumes local/operator resources rather than shared BYOK spend, and the new tests directly exercise both the zero-limit block and the counter-increment-on-success path. The one real wrinkle is that recordVisualVisionUsage is invoked with a hardcoded status of "ok" in both the success path (src/queue/processors.ts, final block) and the no-usable-output/provider-failure path (the `if (!visionText)` block), even though the detail string distinguishes a real failure from a genuine success — only the quota_exceeded call site passes a differentiated status.

Nits — 5 non-blocking

• src/queue/processors.ts — recordVisualVisionUsage is called with the literal status "ok" for both the true-success path and the no-usable-output/provider-failure path (`visionResponse.failure ? ... : "no usable output"` still passes status "ok"); consider passing a distinct status like "no_output" or "provider_error" so any downstream consumer of the status field can't conflate a failed call with a successful one.
• test/unit/visual-vision-wiring.test.ts — the new "adds no finding when the provider returns 200 with no usable text" test doesn't assert the BYOK counter incremented for that branch, leaving the (now status:"ok") no-output recording path's accounting effect unverified even though it's the branch most likely to be miscounted.
• The PR description doesn't cite an issue number this fixes/closes — worth confirming it links to an eligible open issue per repo convention before merge.
• The flagged "secret" at test/unit/visual-vision-wiring.test.ts:241 (`sk-ant-vision-key`) is a test fixture consistent with the pre-existing key literals used elsewhere in this same file, not a real credential.
• Differentiate the status argument passed to recordVisualVisionUsage between the true success path and the no-output/provider-failure path instead of hardcoding "ok" for both (src/queue/processors.ts).

Concerns raised — review before merging

• No linked issue detected — If this PR is intended to solve an issue, link it explicitly in the PR body.
• Maintainer requires a linked issue — Link the relevant issue (for example Closes #123) before opening the PR.
• Possible leaked secret in the diff (generic_secret_assignment) — Remove the secret from the diff, rotate the exposed credential, then re-run the gate.

Signal	Result	Evidence
Code review	❌ 3 blockers	1 reviewer
Linked issue	⚠️ Missing	No linked issue or no-issue rationale found.
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 (no linked issue context).
Validation posture	✅ 25/25	PR body includes validation/test evidence.
Contributor workload	✅ 10/10	Author activity: 48 registered-repo PR(s), 40 merged, 334 issue(s).
Contributor context	✅ Confirmed Gittensor contributor	JSONbored; Gittensor profile; 48 PR(s), 334 issue(s).
Gate result	❌ Blocking	Repo-configured hard blocker found.

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: not available
• Official Gittensor activity: 48 PR(s), 334 issue(s).
• PR-specific overlap: none found.

Contributor next steps

• Treat this as maintainer-lane context rather than normal contributor-lane activity.
• Explain no-issue PR.
• Link the issue being solved, or explicitly explain why this is a no-issue PR.

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