Bulk Issue Solver

Orchestrate parallel mc (mastracode) headless instances to debug and fix multiple GitHub issues simultaneously. You act as the supervisor, spawning workers, monitoring progress, reviewing output, and creating PRs.

Inputs

$ARGUMENTS should be a space-separated list of GitHub issue numbers, e.g. 1234 5678 9012.

If no arguments are provided, use the GH CLI and the user's contribution history to recommend issues:

RUN gh issue list --state open --limit 50 --json number,title,labels,assignees RUN git log --author="$(git config user.name)" --pretty=format:'' --name-only --since="6 months ago" | sed 's|/[^/]*$||' | sort | uniq -c | sort -rn | head -20

Summarize the user's contribution areas and match them to open issues. Ask the user which issues to work on before proceeding.

Setup

For each issue number in the list:

1. Create a git worktree with a dedicated branch:

   git worktree add ../$(basename $PWD)-issue-<NUMBER> -b fix/issue-<NUMBER>
   

2. Install and build in the worktree. Run 2 worktrees at a time to manage CPU:

   cd ../$(basename $PWD)-issue-<NUMBER> && pnpm i && pnpm build
   

3. Spawn an mc headless instance in each worktree with a generous timeout (30 minutes):

   cd ../$(basename $PWD)-issue-<NUMBER> && pnpx tsx <path-to-mastracode>/src/main.ts --timeout 1800 --prompt "Activate the understand-issue skill for issue <NUMBER>"
   

   Run all instances as background processes with a matching timeout on execute_command. Track each PID.

Monitoring

Create a reports/ directory in the main project root. For each issue, maintain a reports/issue-<NUMBER>.md file with:

• Issue number, title, and link
• Current status (Analyzing / Implementing / Tests passing / PR open / Done)
• Summary of the approach and changes
• PR URL once created
• Any blockers or notes

Check all processes every 3 minutes. For each check: