make-blog-post
Pass
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [SAFE]: The skill incorporates a robust set of security and content rules that forbid the inclusion of API keys, tokens, or private data, and enforces a mandatory human approval step before any files are written to disk.
- [INDIRECT_PROMPT_INJECTION]: The skill exposes an attack surface by ingesting external data from URLs and local files for voice analysis. It mitigates this risk with explicit instructions to treat samples as style guides only and to ignore any embedded commands within them.
- [EXTERNAL_DOWNLOADS]: The skill identifies the need to fetch content from user-provided URLs to perform writing-style analysis. This is a functional requirement and is handled as a data retrieval task rather than remote code execution.
Audit Metadata