resource-safe-generation
Pass
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute local Python scripts (
../../scripts/llm_runtime.pyandscripts/resource_handoff.py) to manage model lifecycles and verify the locality of the LLM endpoint. - [INDIRECT_PROMPT_INJECTION]: The skill establishes an automated feedback loop where results and blockers from the offline generation phase are written to
RESOURCE_RESUME.mdto be processed by the agent in the next model turn. - Ingestion points:
RESOURCE_RESUME.mdfile content processed at the start of a resumed session. - Boundary markers: None specified in the instructions.
- Capability inventory: Local Python script execution, file system read/write, and local network requests to the ComfyUI HTTP API.
- Sanitization: No explicit sanitization of the resume data is described, though the data is generated by the skill's own local runner process.
Audit Metadata