reviewing-cryptography
Pass
Audited by Gen Agent Trust Hub on Jul 30, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill documentation suggests using defuddle.md to fetch and convert external web pages to Markdown. The instructions include a clear warning that this action sends the target URL to a third-party service, which may leak engagement details.- [COMMAND_EXECUTION]: The skill provides numerous patterns for using command-line utilities to perform security reviews, including ripgrep (rg) for finding code flaws, gitleaks for secret detection, and networking tools such as nmap, testssl.sh, and sslyze.- [PROMPT_INJECTION]: The skill creates a surface for indirect prompt injection by fetching and processing external content from URLs via an extraction service.
- Ingestion points: The skill uses curl to fetch content from defuddle.md based on provided URLs.
- Boundary markers: There are no explicit instructions to wrap or delimit the external content to prevent the agent from obeying embedded instructions.
- Capability inventory: The skill context includes access to file searching (rg), secret scanning (gitleaks), and network scanning tools.
- Sanitization: The external service provides text extraction but does not perform security-based sanitization or filtering of the content for malicious instructions.
Audit Metadata