Skip to content

AGENTS.md

Latest commit

 

History

History
68 lines (51 loc) · 1.84 KB

AGENTS.md

File metadata and controls

68 lines (51 loc) · 1.84 KB

AGENTS.md - AI Agent Workflow Guide

Configuration for AI agents working with the mgrep codebase.

For full project documentation, see README.md For development patterns, see CLAUDE.md

Quick Reference

npm run build          # Build TypeScript
npm run test           # Run all tests
npm run test:unit      # Vitest unit tests
npm run start -- mcp   # Start MCP server

Issue Tracking (Beads)

This project uses bd (beads) for issue tracking.

bd prime                              # Get workflow context
bd ready                              # Find unblocked work
bd create "Title" --type task -p 2    # Create issue
bd close <id>                         # Complete work

For full workflow details: bd prime

MCP Development

When modifying MCP tools:

  1. Add/update tests in:

    • src/commands/watch_mcp.test.ts - Unit tests
    • src/commands/watch_mcp.helper.test.ts - Helper functions
    • src/commands/watch_mcp.integration.test.ts - Integration tests

  2. Run tests before committing:

    npm run test:unit -- watch_mcp

  3. Key files:

    • src/commands/watch_mcp.ts - MCP server
    • src/lib/test-mcp-client.ts - TestMCPClient
    • src/lib/store.ts - TestStore

Session Completion Protocol

When ending a work session, complete ALL steps:

  1. File issues for remaining work: bd create
  2. Run quality gates: npm run lint && npm run test
  3. Close finished work: bd close <id>
  4. Push to remote: 
    git pull --rebase
bd sync
git push
git status  # MUST show "up to date with origin"
  5. Verify: All changes committed AND pushed

Critical Rules:

  • Work is NOT complete until git push succeeds
  • NEVER stop before pushing - that leaves work stranded locally
  • If push fails, resolve and retry until it succeeds