feat(slop): BYOK frontier model for the AI slop advisory

[JSONbored]

What

Extends the AI slop advisory (#724) to optionally use the maintainer's own frontier model (Anthropic/OpenAI) instead of free Workers AI — reusing the existing per-repo encrypted key + aiReviewByok flag (no new key or setting). One BYOK key serves both AI features. Advisory-only either way; BYOK never changes whether the slop pass can block (it can't).

How

• ai-review.ts: extract callAiProvider (shared BYOK HTTP/timeout/error handling) — runProviderReview now wraps it. Export ProviderFailure + DEFAULT_BYOK_DAILY_REPO_LIMIT.
• repositories.ts: generalize countByokAiReviewEventsForRepoSince → countByokAiEventsForRepoSince — one shared per-repo/day BYOK cap across all AI features (review + slop), so enabling more capabilities can't multiply frontier spend.
• ai-slop.ts: the providerKey path skips the free neuron budget, checks the shared BYOK cap, calls callAiProvider + parseSlopOpinion, and records byok:<provider> usage. Workers-AI fallback unchanged. Fail-safe on every path.
• processors runAiSlopForAdvisory: decrypt the repo key (a declared provider override must match the stored key) and pass it through.

Tests

BYOK provider path (provider used, Workers AI not) + the shared BYOK cap quota; existing Workers-AI gating/fail-safe still green. Typecheck + UI lint + OpenAPI clean; 97% coverage held.

[ghost]

Note

Gittensory Gate skipped

PR closed before full evaluation. No late first comment was created.

Signal	Result	Evidence	Action
Gate result	⚠️ Skipped	#728 is no longer open.	No action.

---

💰 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.

[superagent-security]

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

[ghost]

gittensory · advisory review

Reviewed 5 changed file(s) — two independent AI reviewers.

Suggested action: 🛠️ Request changes. (reviewers split: request changes / merge) Address the suggestions below before merging.

Reviewer A · gpt-oss-120b — recommends 🛠️ request changes
The PR adds BYOK (maintainer‑paid) AI support to the slop advisory flow, shares counting logic with the review path, and updates related DB queries and tests. The intent is clear and the implementation largely follows the existing patterns.

Suggestions

• Import RepositorySettings (and getDecryptedRepositoryAiKey if not already imported) at the top of src/queue/processors.ts to satisfy TypeScript.
• Verify that no other file still imports the old countByokAiReviewEventsForRepoSince name; replace any leftover imports with the new countByokAiEventsForRepoSince.
• Consider adding a unit test for the case where providerKey is supplied but the stored key’s provider mismatches, ensuring the fallback to Workers‑AI is exercised.

Worth double-checking

• Missing imports could cause the build to fail, preventing deployment.
• Changing the DB query to count all BYOK events removes the feature filter; ensure no downstream logic relied on that filter.

Reviewer B · nemotron-3-120b-a12b — recommends ✅ merge
The PR adds BYOK (Bring Your Own Key) support for the AI slop advisory feature, allowing maintainers to use their own frontier models for generating slop advisories while respecting a shared per-repo/day BYOK quota across all AI features. It refactors shared BYOK logic between ai-review and ai-slop services, updates repository functions to count BYOK events universally, and includes comprehensive tests. The changes maintain the public/private boundary by keeping advisory generation internal and only outputting sanitized findings.

Suggestions

• Consider adding a comment in src/services/ai-slop.ts explaining why freeCalls is set to 0 for BYOK (to clarify it doesn't consume the free neuron budget).
• In test/unit/ai-slop.test.ts, the new BYOK usage test could assert that the fetched model matches the stored key's model (if specified) for stronger validation.
• Ensure the DEFAULT_BYOK_DAILY_REPO_LIMIT export in ai-review.ts is actually used elsewhere (it is, via env var fallback) - no action needed.

Worth double-checking

• The shared BYOK daily repo limit means enabling both AI review and slop advisory will consume from the same quota - verify this aligns with product expectations (it appears intentional).
• The BYOK key decryption path (getDecryptedRepositoryAiKey) relies on existing encryption infrastructure; confirm key rotation/secrets management is handled elsewhere (pre-existing).
• Unit test for BYOK cap seeds a prior event but doesn't verify the exact count logic - though the test passes, adding an assertion on byokUsed would increase confidence.