The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/skill-map.py executes the ripgrep (rg) utility using subprocess.run. This is used to perform high-performance file searches for skill definitions and references. The implementation follows security best practices by using argument lists rather than shell strings and enforcing strict validation on skill names via regular expressions.\n- [PROMPT_INJECTION]: The skill scans local files to discover skill references, creating an ingestion surface for potentially untrusted data.\n
Ingestion points: The discover_skills and search_pattern functions in scripts/skill-map.py read content from local files throughout the scanned directories.\n
Boundary markers: The skill instructions and documentation (references/ignore-policy.md) explicitly define a default ignore policy for high-volume and sensitive directories (e.g., VCS, build outputs, and agent state files).\n
Capability inventory: The skill possesses the capability to execute the rg command and read local file contents.\n
Sanitization: User-provided skill names are validated against a restricted regular expression (^[a-z0-9](?:[a-z0-9-]{0,62}[a-z0-9])?$) before being processed.

skill-map