GitHub Issues

Development

Auto-fix GitHub issues with parallel AI agents

What You Can Do

  • Fetch issues by filter — Query by label, milestone, assignee, state, or repository
  • Parallel sub-agents — Spawn up to 8 concurrent agents for simultaneous fixes
  • Auto-branch and PR — Automatically create branches, implement fixes, and open pull requests
  • Fork mode — Work on forked repos with proper upstream tracking
  • Watch mode — Continuous polling at configurable intervals for new issues
  • PR review handler — Detect actionable review comments and spawn fix agents
  • Cron mode — Sequential processing for scheduled CI/CD automation
  • Dry-run — Preview what would happen without executing fixes
  • Telegram notifications — Real-time updates on progress and completions
  • 6-phase pipeline — Fetch, filter, pre-flight, spawn, collect results, handle reviews

    • Try Asking

  • "Fix all issues labeled 'bug' in this repo"
  • "Watch for new issues every 10 minutes and auto-fix them"
  • "Process issues in milestone v2.0 assigned to me"
  • "Handle review comments on open PRs"
  • "Dry-run fixes for the last 5 issues"
  • "Auto-fix issues using fork mode against upstream"

    • Pro Tips

  • Watch mode creates a continuous fix loop — great for active repositories
  • Parallel mode (up to 8 agents) dramatically speeds up multi-issue batches
  • Dry-run mode previews everything before making any changes
  • Fork mode is essential for open-source contributions
  • Cron + reviews-only mode handles PR feedback automatically
  • Confidence check (7/10 minimum) prevents low-quality fixes from being submitted