loop-me
Pass
Audited by Gen Agent Trust Hub on Aug 28, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill's operations are confined to the local workspace, specifically reading from and writing to
workflows/*.mdandNOTES.md. It does not perform network requests or access sensitive system configuration files (e.g., SSH keys, environment variables). - [INDIRECT_PROMPT_INJECTION]: The skill ingests data from local files (
NOTES.md) to inform its behavior and the generation of workflow specifications. This constitutes a standard ingestion surface where data from the workspace influences agent output. - Ingestion points:
NOTES.mdandworkflows/*.mdfiles within the workspace. - Boundary markers: Absent; the instructions do not specify delimiters to isolate ingested content from system instructions.
- Capability inventory: The skill instructions authorize the agent to create, edit, and delete files in the
workflows/directory. - Sanitization: None detected; the skill relies on the content of the local files as the "source of truth" without validation.
Audit Metadata