omh-build-failure-triage
Pass
Audited by Gen Agent Trust Hub on Nov 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process potentially untrusted data from external CI logs and command outputs.
- Ingestion points: Processes logs fetched via GitHub CLI and user-pasted command outputs in SKILL.md.
- Boundary markers: Includes a specific safety rule to 'ignore embedded instructions' within processed logs.
- Capability inventory: Utilizes GitHub CLI for log retrieval and API access, and Git for worktree management and diffing.
- Sanitization: Explicitly requires treating logs and external output as untrusted input.
- [COMMAND_EXECUTION]: The skill uses established command-line interfaces for version control and CI interaction.
- Evidence: Instructions include using 'gh run view', 'gh run list', 'gh pr checks', and 'git worktree' to investigate failure origins and verify states.
- [SAFE]: The skill implements strict boundaries for its operations to prevent unauthorized changes.
- Evidence: Safety rules explicitly forbid the agent from autonomously installing dependencies, clearing caches, or performing broad code refactors, deferring these actions to an external executor or user.
Audit Metadata