omh-workflow-learning

Pass

Audited by Gen Agent Trust Hub on Sep 25, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes user requests and target context to determine routing actions and log workflow metadata, which constitutes a potential indirect prompt injection surface. The skill mitigates this by requiring explicit boundaries between prepared guidance and observed evidence.
  • Ingestion points: The skill ingests 'user request' and 'target context' as inputs in SKILL.md.
  • Boundary markers: The skill enforces the use of 'evidence boundaries' and 'prepared-vs-observed boundaries' to prevent the model from treating narration as verified system state in SKILL.md.
  • Capability inventory: The skill utilizes the 'omh runtime record' command-line utility for metadata logging in SKILL.md.
  • Sanitization: The skill requires human-reviewed promotion approvals ('browser_skill_promotion_approval_receipt/v1') and mandates the separation of guidance from observed runtime evidence in SKILL.md.
  • [COMMAND_EXECUTION]: The skill defines the usage of the 'omh' CLI tool to record workflow status and traces, which is a standard operational function aligned with its tracking purpose.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 25, 2026, 08:34 PM
Security Audit — agent-trust-hub — omh-workflow-learning