---

name: context-map description: Fast repository indexing for RLM workflows. Use before analysis to reduce context drift.

Context Map Skill

Create a lightweight index of the repository structure without reading full file contents.

Usage

Copycontext-map [path]   # Generate index for path (default: current repo)

Output Format

Copy{
  "generated_at": "2024-01-15T12:00:00Z",
  "entrypoints": ["app/main.py", "cli.py"],
  "core_modules": ["domain/", "services/", "infra/"],
  "tests": ["tests/"],
  "hotspots": ["billing/", "auth/"],
  "config_files": ["pyproject.toml", ".env.example"],
  "commands": {
    "format": "uv run ruff format .",
    "lint": "uv run ruff check .",
    "test": "uv run pytest -v",
    "typecheck": "uv run ty check"
  }
}

Mapping Process

1. Structure Scan

• List top-level directories
• Identify source, tests, config, docs

2. Entry Point Detection

• Find main.py, app.py, cli.py
• Find __main__.py in packages
• Check pyproject.toml for scripts

3. Hotspot Identification

• Modules with most recent changes (git log)
• Modules with most files/lines
• Commonly referenced modules (grep imports)

4. Command Detection

• Parse pyproject.toml (scripts, taskipy)
• Parse Makefile targets
• Parse package.json scripts

Integration with Pipeline

Use at Step 00 (Init) or Step 01 (Analyze):

Copy## Before Analysis

1. Run `context-map` on repo root
2. Use output to guide MAP phase
3. Focus DRILL on hotspots + relevant modules

Best Practices

• Generate once per session, refresh if changing branches
• Use as input for step-01-analyze.md
• Combine with explorer skill for deeper dives