friction-to-proof
Pass
Audited by Gen Agent Trust Hub on Oct 6, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [SAFE]: The skill instructions and referenced files were analyzed, and no malicious patterns, prompt injections, or security risks were identified.
- [NO_CODE]: The skill is composed exclusively of Markdown documentation and procedural guidelines; it does not ship with any Python, JavaScript, or shell scripts.
- [EXTERNAL_DOWNLOADS]: In
references/research-basis.md, the skill lists several URLs pointing to academic papers and professional resources from trusted and well-known domains, including harvard.edu, springer.com, sagepub.com, google.com, and ncbi.nlm.nih.gov. These are static references used for theoretical context and do not represent a remote code execution or data exfiltration risk. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to process 'pissed-off' user input (rants), which is a common vector for indirect prompt injection. However, the protocol mitigates this by requiring the agent to translate the input into structured artifacts (Decision Challenge Cards) and explicitly 'strip contempt' and 'translate character judgments' before outputting. Given the skill's lack of autonomous capabilities or access to sensitive system tools, the risk of an injection attack causing actual harm is negligible.
Audit Metadata