The Agent Skills Directory

[COMMAND_EXECUTION]: The file scripts/auto-update.py uses subprocess.run to call a sibling Python script analyze_commands.py. This is an intended management function for the skill to regenerate its local command index files by scanning the project's directory structure.
[DATA_ACCESS]: The scripts/analyze_commands.py script reads markdown files from the .claude/commands, .claude/agents, and .claude/skills directories to extract metadata (frontmatter) for indexing. This data is stored locally in JSON files for the skill's own lookups and is not transmitted externally.
[INDIRECT_PROMPT_INJECTION]: The skill builds its command database (command.json) by ingesting content from various markdown files in the workspace, which represents an attack surface for indirect prompt injection if those files are from untrusted sources.
Ingestion points: The analyze_commands.py script reads from the commands/, agents/, and skills/ directories.
Boundary markers: None identified in the provided indexing or processing logic.
Capability inventory: The skill uses tools like Read, Grep, and Glob and can orchestrate command execution sequences.
Sanitization: The indexing script extracts raw frontmatter values without specific sanitization filters.

ccw-help