agent-loadout
Pass
Audited by Gen Agent Trust Hub on Sep 7, 2026
Risk Level: SAFEDYNAMIC_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [DYNAMIC_EXECUTION]: The helper scripts perform dynamic imports of Node.js modules from the host Pi package directory using paths computed at runtime. This allows the skill to utilize the host platform's internal utilities while ensuring that execution is scoped to the verified installation directory of the Pi agent.
- [COMMAND_EXECUTION]: The skill facilitates the execution of local Node.js scripts to apply configuration changes. It passes user-defined settings as JSON-encoded arguments to these scripts, which then perform validated file operations.
- [INDIRECT_PROMPT_INJECTION]:
- Ingestion points: The skill reads existing configuration data from
settings.jsonfiles at the user and project levels. - Boundary markers: The skill treats settings as structured JSON and enforces strict schema validation for agent names, model identifiers, and thinking levels.
- Capability inventory: The skill can modify configuration files and execute its own Node.js-based helper scripts.
- Sanitization: Input validation logic in
normalizePolicyandvalidateRequestsprevents the injection of arbitrary instructions into persistent settings fields.
Audit Metadata