aif
Pass
Audited by Gen Agent Trust Hub on May 28, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: Executes a local security scanner (
security-scan.py) to analyze any external skills for malicious content before they are integrated into the project. - [COMMAND_EXECUTION]: Uses a specialized Node.js script (
update-config.mjs) to manage YAML configuration changes, providing a safer alternative to free-form text generation by the agent. - [EXTERNAL_DOWNLOADS]: Fetches and installs additional agent skills from the skills.sh registry based on detected project needs.
- [INDIRECT_PROMPT_INJECTION]: Reads project dependency files (e.g., package.json, composer.json, requirements.txt) to determine the technology stack.
- Ingestion points: SKILL.md (Step 1) reads multiple dependency files including package.json, composer.json, requirements.txt, pyproject.toml, go.mod, Cargo.toml, and docker-compose.yml.
- Boundary markers: No explicit markers or "ignore instructions" blocks are used when reading these files.
- Capability inventory: Includes Bash (mkdir, rm, npx, python, node) and Write tools.
- Sanitization: The skill relies on tech-stack pattern matching; no explicit input sanitization of file contents is documented.
Audit Metadata