pyhd
Pass
Audited by Gen Agent Trust Hub on Mar 5, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: Uses
uv run rufffor file formatting and linting. It also executesuv run python -cto retrieve documentation using thehelp()function. These are standard developer operations. - [PROMPT_INJECTION]: The skill has an indirect prompt injection surface as it reads local code and external search results.
- Ingestion points: Local files and
google_web_searchresults. - Boundary markers: None identified.
- Capability inventory: Includes file modification (
smart_edit,replace) and command execution throughuv. - Sanitization: No explicit content filtering or instruction scrubbing for external data is mentioned.
Audit Metadata