lattice-rest-integration
Pass
Audited by Gen Agent Trust Hub on Aug 4, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONCREDENTIALS_UNSAFEDATA_EXFILTRATION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill fetches an OpenAPI specification from the official vendor domain (developer.anduril.com) to ensure the integration uses correct API paths.\n- [COMMAND_EXECUTION]: Instructions direct the agent to install language-specific SDKs if they are not present and to execute the generated integration to validate its functionality against a sandbox environment.\n- [CREDENTIALS_UNSAFE]: The skill provides guidance on authenticating to the Lattice Sandbox by reading tokens from environment variables or .env files, which is standard practice for developer tools.\n- [DATA_EXFILTRATION]: The skill involves sending data to the Lattice Sandbox API (Entities, Tasks, and Objects) as part of the integration workflow.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes data from external sources, including the vendor's OpenAPI spec and API responses from the Sandbox environment. Ingestion points are the remote specification file and Sandbox API endpoints. There are no explicit boundary markers or sanitization routines defined for these inputs, which interact with sensitive capabilities like package installation and code execution.
Audit Metadata