api-patterns
Pass
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: LOWCOMMAND_EXECUTION
Full Analysis
- [Indirect Prompt Injection] (LOW): The skill processes external codebase data through its
api_validator.pyscript, which introduces a potential surface for processing untrusted data. - Ingestion points: The
<project_path>argument passed to the Python script. - Boundary markers: None identified in the SKILL.md definition.
- Capability inventory: The agent has access to standard file system tools (
Read,Write,Edit,Glob,Grep) and can execute Python scripts. - Sanitization: Not specified in the skill metadata; security relies on the internal implementation of the validation script.
- [Command Execution] (LOW): The skill instructs the agent to execute a local Python script (
scripts/api_validator.py). While this is standard functionality for many agent skills, it constitutes dynamic code execution on the host environment.
Audit Metadata