The Agent Skills Directory

[SAFE] (SAFE): The skill provides an extensive framework for managing agent skills with a strong emphasis on security and best practices.
[EXTERNAL_DOWNLOADS] (LOW): The skill documentation describes workflows for installing external code using npx skills add and skill-manager install.
Evidence: Found in references/finding-skills.md and references/skill-manager-usage.md.
Mitigation: The skill includes a 'Security Audit' checklist (Category 2, 4, and 8 mitigation) in references/finding-skills.md that explicitly instructs the agent/user to scan for prompt injection, data exfiltration, and supply chain risks before installation.
[COMMAND_EXECUTION] (LOW): The skill references a skill-manager script that performs file system operations (create, remove, symlink) to manage the skill lifecycle.
Evidence: Found in references/skill-manager-usage.md. The operations are restricted to the local user's skill directories (e.g., ~/.claude/skills/).
[PROMPT_INJECTION] (SAFE): The skill contains defensive instructions to prevent the agent from being 'persuaded' to skip steps or ignore rules (specifically for TDD).
Evidence: references/testing-methodology.md uses psychological principles (Cialdini) to ensure agent adherence to safety and quality protocols, acting as a counter-injection measure.

skills-management