The Agent Skills Directory

[COMMAND_EXECUTION]: Executes repository-specific scripts such as python bootstrap/scripts/update_skills_mirror.py and standard maintenance commands like python -m compileall.
[EXTERNAL_DOWNLOADS]: Invokes npx -y skills add . --list which downloads and executes the 'skills' package from the npm registry at runtime.
[PROMPT_INJECTION]: The skill facilitates the ingestion of content from external repositories, which creates a potential surface for indirect prompt injection attacks.
Ingestion points: Clones and imports data from external repositories into the .references/ and canonical skill directories (SKILL.md).
Boundary markers: None explicitly defined to delimit imported content from system instructions.
Capability inventory: Includes shell command execution via Python/npx and significant file-system write access across the repository structure.
Sanitization: Relies on manual instructions to "adapt" content and "remove upstream extras" rather than automated sanitization.
[SAFE]: References access to configuration files located at ~/.agents/config.toml and $HUMAN_MATERIAL_PATH/.agents/config.toml for the purpose of establishing configuration precedence.

uv-bootstrap-skill-maintenance