fix(knowledge): preserve live document processing during recovery

[waleedlatif1]

Summary

• Extend the existing recovery guard with bounded database and cancellable Trigger checks, preserve work when status is unknown, and recheck the full observed generation before resetting it.
• Apply the same protection to scheduled recovery, connector sweeps, manual retry, and legacy queue redelivery.
• Add sanitized OCR response and Google token-transport diagnostics without changing provider retries, permission checks, or public errors.

Type of Change

• Bug fix

Testing

• 912 focused unit tests and 44 disposable PostgreSQL/Redis integration tests passed, including queue delays, generation races, legacy upload redelivery, durable delivery, lifecycle guards, SDK request cancellation, database acquisition deadlines, and diagnostic redaction.
• Full monorepo type-check, lint and lint:check, 47 audits, API boundary validation, block registry and docs-manifest checks passed.
• Cleanup and final independent reviews completed. No schema migration is required. Live verification after deployment remains necessary; this does not reset exhausted attempts or resolve provider access and database infrastructure failures.

Checklist

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

[vercel]

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

1 Skipped Deployment

Project	Deployment	Actions	Updated
docs	Skipped		Sep 19, 2026 12:25am UTC

[greptile-apps]

Confidence Score: 5/5

The PR appears safe to merge; the previously reported regressions are resolved and no new actionable failure remains.

Summary

This PR protects live or indeterminate document-processing generations from being replaced during recovery and applies that protection consistently across scheduled, connector, manual retry, and legacy redelivery paths. It also makes worker outcomes explicit and adds sanitized provider diagnostics.

• Rechecks durable outbox and Trigger.dev liveness before replacing a processing generation.
• Uses snapshot guards and bounded inspection deadlines to prevent races with concurrent lifecycle transitions.
• Preserves live work through connector sweeps, manual retries, and legacy outbox redelivery.
• Distinguishes indexed documents from harmless unavailable, unclaimed, or superseded task outcomes.
• Logs bounded, allowlisted OCR diagnostics without exposing provider messages or document contents.

Diagram

%%{init: {'theme': 'neutral'}}%%
flowchart TD
  A[Recovery candidate] --> B{Active outbox carrier?}
  B -->|Yes| C[Preserve generation and defer recovery]
  B -->|No| D{Active Trigger.dev run?}
  D -->|Yes| C
  D -->|Lookup unavailable| E[Preserve generation and schedule short recheck]
  D -->|No active work| F[Recheck document snapshot]
  F -->|Snapshot changed| G[Leave winning generation untouched]
  F -->|Snapshot unchanged| H[Adopt or replace generation]
  H --> I[Enqueue durable processing event]
  I --> J[Process document]
  J -->|Index committed| K[Return indexed outcome]
  J -->|Unavailable, unclaimed, or superseded| L[Return harmless skipped outcome]

Loading

_{Reviews (4) · Last reviewed commit: "fix(knowledge): bound recovery database ..."}

[cubic-dev-ai]

All reported issues were addressed across 16 files

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

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

[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]

All reported issues were addressed across 16 files

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

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

[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]

All reported issues were addressed across 17 files

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

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

[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 17 files

Confidence score: 5/5

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

_{Re-trigger cubic}