omh-github-event-ops
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external GitHub event data (issues, pull requests, CI events) which can be controlled by third parties. Malicious actors could embed instructions within PR descriptions or issue comments to attempt to influence the agent's behavior during triage or review tasks.
- Ingestion points: GitHub event data (PRs, issues, CI events) referenced in SKILL.md.
- Boundary markers: The skill instructions emphasize a "prepared-vs-observed boundary" to distinguish between agent guidance and actual platform evidence.
- Capability inventory: Shell command execution via the 'omh' CLI tool as seen in SKILL.md.
- Sanitization: No explicit sanitization or filtering logic for external content is defined within the provided skill instructions.
- [COMMAND_EXECUTION]: The skill utilizes the 'omh' command-line interface to record runtime status ('omh runtime record'). This command is part of the specific framework's workflow tracking and achievement system.
Audit Metadata