atai-manual-generation-agent
Pass
Audited by Gen Agent Trust Hub on Oct 31, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is a well-documented integration for the Archetype AI platform. It provides a utility script (
run_mga_agent.py) that uses the officialarchetypeaiPython SDK to interact with managed MGA blueprints. The analysis did not detect any malicious patterns such as prompt injection, unauthorized data exfiltration, or obfuscated code. - [DATA_EXPOSURE]: The skill correctly manages sensitive API credentials by using environment variables (via a
.envfile) rather than hardcoding them, aligning with security best practices. - [EXTERNAL_DOWNLOADS]: The skill references and uses the official
archetypeaiPython package and connects to the official vendor infrastructure (archetypeai.app). These are verified vendor resources and do not pose a security risk. - [INDIRECT_PROMPT_INJECTION]: As the agent processes video and audio data to generate manuals, it is theoretically susceptible to indirect prompt injection if a video contains adversarial audio or visual instructions. However, the skill implements specific prompting strategies (e.g., instructing the model not to invent steps and to rely only on visible/spoken evidence) that serve as mitigations. Given the intended use case of processing user-supplied procedure videos, this remains a low-risk architectural factor rather than a specific vulnerability.
Audit Metadata