omh-research-department

Pass

Audited by Gen Agent Trust Hub on Sep 25, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and organize external data via 'source inboxes' and 'knowledge stores' for research purposes. This ingestion of untrusted external content represents a surface where malicious instructions could be embedded in research materials.
  • Ingestion points: External research sources, source inbox buckets, and knowledge-store content processed in SKILL.md.
  • Boundary markers: The skill explicitly mandates a 'not-evidence boundary' to prevent the agent from treating plans or synthesis as hard evidence, and requires separating observed sources from inference.
  • Capability inventory: Executes shell commands via the omh CLI tool for runtime recording.
  • Sanitization: Relies on the logical separation of findings, notes, and verification needs into distinct buckets to mitigate cross-contamination.
  • [COMMAND_EXECUTION]: The skill utilizes a shell command omh runtime record to log the lifecycle of the research department workflow. This command is a component of the local 'oh-my-hermes' runtime environment mentioned in the metadata and file paths.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 25, 2026, 08:34 PM
Security Audit — agent-trust-hub — omh-research-department