omh-skill
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to manage local skills, which involves reading, writing, and updating manifest files (e.g.,
SKILL.md). This process creates a vulnerability surface where malicious instructions could be embedded in the data (skills) processed by the agent. - Ingestion points: The skill is triggered to list, search, add, remove, or edit skill manifests and directories (SKILL.md).
- Boundary markers: The instructions provide guidance on separating framework logic from executor behavior and emphasize verification, though these are natural language instructions rather than technical delimiters.
- Capability inventory: The skill uses the
omhruntime to record status and is authorized to perform manifest updates and delegate code changes to an executor. - Sanitization: No explicit technical sanitization, schema validation, or escaping of the content within managed skills is described.
Audit Metadata