ooda-loop

Pass

Audited by Gen Agent Trust Hub on Aug 22, 2026

Risk Level: SAFENO_CODE
Full Analysis
  • [NO_CODE]: The skill is entirely composed of Markdown files containing documentation, historical examples, and instructional templates. No scripts, binaries, or automated tool configurations are present.\n- [EXTERNAL_DOWNLOADS]: The documentation contains links to the author's official website (deciqai.com) and references well-known services and organizations, including OpenAI, Anthropic, and Stanford University, for background information and research reports. No automated remote code execution or package installation is performed.\n- [PROMPT_INJECTION]: The 'Adaptive Front Door' section provides benign conversational guidance using [WAIT] markers to manage multi-turn interactions. These instructions are intended for user engagement and do not attempt to bypass safety guidelines or override system behavior.\n- [SAFE]: No indicators of obfuscation, credential theft, persistence mechanisms, or unauthorized file access were found. The skill operates purely as a set of logical prompts for strategic analysis.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 22, 2026, 08:15 AM
Security Audit — agent-trust-hub — ooda-loop