analyzing-user-feedback
Pass
Audited by Gen Agent Trust Hub on Apr 16, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [SAFE]: The skill consists exclusively of markdown documentation and JSON configuration files. No executable scripts, binaries, or active code components are present in the repository.
- [PROMPT_INJECTION]: There are no instructions that attempt to override agent safety protocols, bypass system constraints, or perform prompt leaking. The language used is purely instructional and focused on the stated product discovery purpose.
- [DATA_EXFILTRATION]: No network-enabled commands (e.g., curl, wget) or external communication patterns were detected. The skill instructions explicitly encourage the use of redacted data to prevent the exposure of PII.
- [CREDENTIALS_UNSAFE]: No hardcoded API keys, tokens, or credentials are found in the code or instructions. Placeholders in templates are generic and do not solicit sensitive secrets.
- [COMMAND_EXECUTION]: The skill does not contain any shell commands or patterns that would trigger unauthorized system execution. It lacks the '!' command syntax for dynamic context injection.
- [REMOTE_CODE_EXECUTION]: No remote dependencies or external scripts are fetched or executed. The skill operates entirely within the provided markdown context.
- [EXTERNAL_DOWNLOADS]: The skill does not perform any external downloads. Referenced URLs in metadata point to the official skill source on the refoundai.com domain.
Audit Metadata