The Agent Skills Directory

[SAFE]: The skill instructions and associated reference files do not contain any malicious code, external downloads from untrusted sources, or attempts to harvest sensitive credentials.
[COMMAND_EXECUTION]: The skill recommends using standard system utilities like find and rg (ripgrep) to verify that referenced files exist and to clean up temporary development noise within the local skill directory.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process and modify existing skill files, which is a potential ingestion point for untrusted instructions. 1. Ingestion points: The content of skills being repaired or improved. 2. Boundary markers: No specific delimiters are specified to isolate untrusted content from the agent's core instructions. 3. Capability inventory: The agent has access to file system read and write operations along with local command execution. 4. Sanitization: The skill does not include instructions for sanitizing or validating the content of the skills it processes.

skill-development