agency-anatomy

Agency Anatomy — File Index Generator

Regenerate the .wolf/anatomy.md codebase index.

Context

This skill scans the project directory, counts lines in every source file, estimates token usage (lines x 4), and produces a sorted table so agents know which files are expensive to read.

Instructions

Step 1: Determine Project Directory

1. If $ARGUMENTS is provided, use it as the project directory
2. Otherwise, use the current working directory
3. Validate the directory exists

Step 2: Run the Anatomy Hook

Run the anatomy hook script:

PROJECT_DIR="<project-dir>" bash "${CLAUDE_PLUGIN_ROOT}/hooks/anatomy-hook.sh"

If CLAUDE_PLUGIN_ROOT is not set, locate the hook relative to this skill file.

Step 3: Display Results

1. Read the generated .wolf/anatomy.md
2. Report:
   • Total files indexed
   • Total estimated tokens
   • Top 10 largest files by token count
3. Note any files that exceed 10,000 estimated tokens as "context-heavy"

Step 4: Recommendations

If total tokens exceed 500,000:

• Suggest which directories to .claudeignore
• Recommend splitting large files

If any single file exceeds 20,000 tokens:

• Flag it as a candidate for decomposition