omh-workspace-audit
Pass
Audited by Gen Agent Trust Hub on Nov 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the local repository (files, configurations, and skill definitions) to produce an inventory and risk map.
- Ingestion points: Repository files and directory structure in the workspace root.
- Boundary markers: The skill defines an 'evidence boundary' and 'read-only boundary' to constrain its operations.
- Capability inventory: File reading across the local repository surface.
- Sanitization: Explicit instructions mandate the redaction of secret values and the use of redacted key names in findings.
- [COMMAND_EXECUTION]: The skill includes a hardcoded shell command for tracking runtime status via an internal tool.
- Evidence: omh runtime record --skill workspace-audit --harness workspace-audit --status started (SKILL.md).
Audit Metadata