github-pr-triage

Automate GitHub pull request triage by spawning an independent background task for every open PR, streaming real-time per-PR analysis to surface urgent issues, and concluding with a comprehensive final triage report. This parallel processing approach handles active development repositories efficiently.

Yes, streaming real-time analysis results for each pull request is supported using background_output streaming. As each PR is analyzed independently in the background, live results are surfaced continuously, enabling timely maintenance decisions and automated closes without waiting for the entire batch to finish.

Handling a high volume of open pull requests is best achieved by enforcing a 1-PR-1-TASK policy that analyzes each PR as a separate background task. This method exhaustively collects open PRs with pagination and processes them in parallel to accelerate maintenance workflows.

Yes, automated PR triage supports conservative auto-close decisions by analyzing each pull request independently and identifying auto-close candidates. It surfaces recommendations throughout the streaming process and includes these decisions in the final comprehensive summary report.

Generate a final triage report after the automated analysis of all open pull requests concludes. The system processes every open PR in parallel, streams intermediate results, and then compiles a comprehensive final summary report containing urgent issues, recommendations, and auto-close candidates.

A limitation of using background tasks for parallel pull request analysis is the strict 1-PR-1-TASK policy, which requires managing each PR independently. This approach is designed for large codebases with active development, so smaller repositories may not benefit from the parallel streaming overhead.