The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use local shell commands to generate UUID v4 strings for metadata. References include uuidgen and python3 -c "import uuid; print(uuid.uuid4())" found in references/creating-skill.md and references/skill-contract.md.
[COMMAND_EXECUTION]: The skill documents the use of npx for installing dependencies, specifically using npx skills add to fetch skills from GitHub repositories as seen in the examples in references/skill-contract.md.
[PROMPT_INJECTION]: The skill's primary function involves reading and auditing external markdown files, which introduces a potential for indirect prompt injection if a target skill contains malicious instructions. 1. Ingestion points: Skill manifest (SKILL.md) and reference markdown files in the target directory. 2. Boundary markers: Not explicitly defined; the agent reads files in their entirety to perform audits. 3. Capability inventory: File system read/write, local command execution (python3, uuidgen, npx). 4. Sanitization: None; the agent processes raw markdown to evaluate stylistic and structural compliance.

managing-skills