reviewing-python-libraries
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDYNAMIC_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructs the agent to use
git cloneto download source code from external, user-specified repositories for review inSKILL.md. - [COMMAND_EXECUTION]: The agent executes multiple shell commands via the
uvtool, includinguv run pytestanduv run ruff, to perform checks on the downloaded code as described inCHECKLIST.md. - [DYNAMIC_EXECUTION]: The skill performs dynamic code execution by running the test suite of the cloned library using
uv run pytest, which executes code not provided within the skill itself. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external libraries, creating a vulnerability surface. 1. Ingestion points: The agent reads and analyzes external source code, documentation, and configuration files. 2. Boundary markers: The instructions lack explicit delimiters to differentiate between repository content and the agent's core instructions. 3. Capability inventory: The agent possesses the capability to execute shell commands and interact with the file system. 4. Sanitization: The skill does not perform sanitization of the library content before processing it.
Audit Metadata