ai-chat
Pass
Audited by Gen Agent Trust Hub on Jul 6, 2026
Risk Level: SAFE
Full Analysis
- [DATA_EXPOSURE_AND_EXFILTRATION]: The skill accesses credentials stored in ~/.build-host/credentials.json. This access is limited to the intended platform integration and is supported by clear instructions to the agent to never print secrets or ask the user for keys in source code or chat.\n- [COMMAND_EXECUTION]: The skill facilitates deployment through specific API endpoints like /api/projects/:uuid/deploy and environment variable management. These actions are standard for the skill's purpose and are constrained by safety rules requiring confirmation for destructive operations.\n- [SAFE]: The skill's behavior aligns perfectly with its stated intent. The implementation instructions focus on minimal, durable changes and respect for user project boundaries.
Audit Metadata