fix(knowledge): name embedding quota, key, and deadline failures in knowledge search

[waleedlatif1]

Summary

• Knowledge search now names the failures a caller can act on instead of collapsing them into a generic 500 "Failed to perform vector search":
  • workspace (BYOK) embedding key out of quota → 503, points at Settings > Provider API keys
  • platform embedding key out of quota → 503, does not blame the workspace's key
  • workspace embedding key rejected → 502
  • retrieval deadline → 504 with the existing deadline message
• A quota error raised while a key's quota pause is already open (or on admission refusal) now keeps the BYOK flag. Before, only the first request after exhaustion was attributed to the workspace key; every request during the 5-minute pause looked like a platform failure
• The knowledge tool path still logs projected 5xx errors, so moving these off the unhandled branch doesn't lose server-side visibility
• Every status stays 5xx, so retry and alerting behavior keyed on server errors is unchanged; platform-key rejections and unknown errors still fall through to the existing generic 500

Type of Change

• Bug fix

Testing

• New regression tests, each confirmed red by reverting its guard: quota-pause BYOK attribution (lib/embeddings/client.test.ts), search error policy mapping + quota-check ordering (lib/knowledge/api/route-policies.test.ts)
• vitest over lib/embeddings, lib/internal/knowledge, lib/knowledge/{api,application,documents,search}, app/api/knowledge, tools/knowledge: 127 files, 1,152 tests passing
• bun run type-check, bun run lint, bun run check:audits (52 audits), docs-manifest:check, block-registry check

Checklist

• Code follows project style guidelines
• Self-reviewed my changes
• Tests added/updated and passing (new tests pass the test-audit authoring gate)
• No new warnings introduced
• I confirm that I have read and agree to the terms outlined in the Contributor License Agreement (CLA)

🤖 Generated with Claude Code

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

1 Skipped Deployment

Project	Deployment	Actions	Updated
docs	Skipped		Sep 29, 2026 5:48am UTC

[cubic-dev-ai]

All reported issues were addressed across 6 files

<sub>Reply with feedback, questions, or to request a fix.

Fix all with cubic | Re-trigger cubic</sub>

[greptile-apps]

Confidence Score: 5/5

[Medium risk] Improves error naming in knowledge search embedding failures.

The PR appears safe to merge; no outstanding or new actionable findings remain.

Summary

The PR gives knowledge-search callers specific responses for embedding quota, rejected workspace-key, and retrieval-deadline failures. It also preserves workspace-key attribution while a quota pause is open and retains logging for projected 5xx tool errors.

_{Reviews (3) · Last reviewed commit: "test(embeddings): prove the quota pause ..."}

[waleedlatif1]

@greptile

[waleedlatif1]

@cubic-dev-ai review this PR

[cubic-dev-ai]

@cubic-dev-ai review this PR

@waleedlatif1 I have started the AI code review. It will take a few minutes to complete.

[cubic-dev-ai]

No issues found across 6 files

Confidence score: 5/5

• Automated review surfaced no issues in the provided summaries.
• No files require special attention.

_{Re-trigger cubic}

[waleedlatif1]

@greptile

[waleedlatif1]

@cubic-dev-ai review this PR

[cubic-dev-ai]

@cubic-dev-ai review this PR

@waleedlatif1 I have started the AI code review. It will take a few minutes to complete.

[cubic-dev-ai]

No issues found across 6 files

Confidence score: 5/5

• Automated review surfaced no issues in the provided summaries.
• No files require special attention.

_{Re-trigger cubic}