The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill documentation mentions installation via 'npx skills add', which is a standard procedure for this platform's ecosystem.- [COMMAND_EXECUTION]: The project includes a Node.js CLI tool in 'cli/src/index.js'. This tool performs legitimate operations: reading the local 'SKILL.md' file and making requests to the official OpenAI and Anthropic API endpoints using environment-provided keys.- [DATA_EXPOSURE]: The skill accesses environment variables for AI service API keys. The 'SECURITY.md' file provides appropriate guidance on handling PII and sensitive data when using LLM-based services.- [PROMPT_INJECTION]: The instructions are focused on persona simulation and 'The Mom Test' methodology. No malicious injection or override patterns were detected.- [INDIRECT_PROMPT_INJECTION]: The skill processes external data such as interview transcripts. However, the 'SKILL.md' contains specific logic ('Truth Filter') to sanitize and filter this input, which mitigates the risk of the model following instructions embedded in that data.

personatwin