animate-objects
Pass
Audited by Gen Agent Trust Hub on Oct 11, 2026
Risk Level: SAFECOMMAND_EXECUTIONPERSISTENCEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to create autonomous logic that reacts to changes in external data sources (CSV, YAML, JSON, API polls). This creates a surface where malicious data in those files could influence the agent's logic.
- Ingestion points: External files, documents, or API responses identified in Step 2 and 3 of the procedure.
- Boundary markers: Step 1 requires restating "no-touch boundaries" and Step 5 requires documenting what the artifact must NOT do.
- Capability inventory: The skill can implement script loops, cron jobs, and file watchers that can modify artifacts or trigger system alerts (SKILL.md, Step 6 and 8).
- Sanitization: Not explicitly mentioned, though Step 6 emphasizes using the "simplest trigger + action loop" and "minimal agency."
- [COMMAND_EXECUTION]: The procedure instructs the agent to implement automation using system tools like cron, GitHub Actions, or custom scripts. The risk is managed through a requirement for explicit user confirmation before taking any action that modifies access or system boundaries (SKILL.md, Step 8).
- [PERSISTENCE]: The core purpose of the skill is to establish ongoing, autonomous automation (persistence) through triggers like file watchers and scheduled tasks. To mitigate risk, the skill mandates a documented "kill switch" or revert path to return artifacts to an inert state (SKILL.md, Deliverables).
Audit Metadata