omh-agent-instructions
Pass
Audited by Gen Agent Trust Hub on Nov 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from repository files to update instructions, which could lead to malicious instruction execution if an attacker can influence those files.
- Ingestion points: The skill reads various instruction files including
AGENTS.md,CLAUDE.md,GEMINI.md,.cursorrules, and.github/copilot-instructions.mdas specified inreferences/instruction-file-method.md. - Boundary markers: The skill uses explicit HTML-style comments (
<!-- omh:agent-instructions:begin -->and<!-- omh:agent-instructions:end -->) to delimit its output and preserve hand-written text. - Capability inventory: The skill is capable of writing to the filesystem (to update the instruction files) and executing build, test, lint, and regeneration commands identified within the repository.
- Sanitization: The skill contains explicit safety rules to prevent the leakage of secrets, tokens, or private hostnames, and it refuses to record volatile data like line numbers or test counts that could lead to drift.
Audit Metadata