stelow-product-discovery
Warn
Audited by Gen Agent Trust Hub on Jul 18, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill's tool references (found in
references/cli-tools/) instruct the agent to install and execute various third-party packages from NPM and GitHub. Key examples includenpx -y @earendil-works/pi-agent-browser,pi install git:github.com/cursor/plugins,pi install git:github.com/PriNova/pi-agent-codebase-workflows, and@plannotator/pi-extension. These sources are not verified by the internal trusted vendor list and execute code directly in the agent's environment. - [COMMAND_EXECUTION]: The skill frequently uses the
bashtool to perform complex management tasks, such as atomic file locking usinglnandsha1sum, and reading workflow configurations via inline Node.js commands (node -e). While these are functional for the framework's state management, they provide a broad surface for command-line interaction. - [SAFE]: No evidence of malicious prompt injection, credential harvesting (e.g.,
.envor.sshaccess), or data exfiltration to unauthorized external domains was detected in the skill's instructions or scripts.
Audit Metadata