aip-setup
Pass
Audited by Gen Agent Trust Hub on Oct 30, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided text ('goal') and passes it to an external tool to initiate video projects.\n
- Ingestion points: User-provided goal/prompt in SKILL.md (Step 5).\n
- Boundary markers: Absent; instructions specify using the user's intent 'in their own words.'\n
- Capability inventory: Calls start_project_from_attachment and handles video file objects.\n
- Sanitization: Absent; the skill relies on the downstream service to validate input content.\n- [COMMAND_EXECUTION]: The skill instructs the agent to guide the user in running platform-native authentication commands (e.g., 'codex mcp login ai-producer', '/mcp'). These are legitimate platform mechanisms for connecting Model Context Protocol (MCP) servers and do not represent unauthorized command execution.
Audit Metadata