oma-explanation
Pass
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data including code diffs and PR descriptions which could contain malicious instructions designed to influence the agent.
- Ingestion points: The skill reads external data via
git diffandgh pr diff(PR content). - Boundary markers: The instructions explicitly direct the agent to treat diff content and PR descriptions strictly as DATA and to ignore any instructions embedded within them.
- Capability inventory: The skill can read repository files, execute
git/ghcommands, and write HTML artifacts to the.agents/results/explain/directory. - Sanitization: The skill employs a structured draft format and an internal renderer (
oma explain render) to generate HTML, which limits the risk of arbitrary script injection into the final document. It also implements pre-generation and post-generation secret gates to prevent the leakage of credentials.
Audit Metadata