pi-setup
Fail
Audited by Gen Agent Trust Hub on May 19, 2026
Risk Level: HIGHCOMMAND_EXECUTIONCREDENTIALS_UNSAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses shell commands to scan the user's live environment and all common shell startup files (~/.zshrc, ~/.bashrc, ~/.profile, etc.) to locate and read API keys.
- [CREDENTIALS_UNSAFE]: Accesses sensitive files including ~/.deepseek and multiple shell RC files to retrieve or verify the presence of API tokens (DEEPSEEK_API_KEY, LING_API_KEY).
- [EXTERNAL_DOWNLOADS]: Triggers the installation of 21 third-party NPM packages and one GitHub repository (fluxgear/pi-thinking-steps) that are not from recognized trusted organizations.
- [REMOTE_CODE_EXECUTION]: Configures the target agent to execute arbitrary shell commands dynamically (using !awk and !zsh -ic) from within its JSON configuration files to retrieve secrets at runtime.
Recommendations
- AI detected serious security threats
Audit Metadata