omh-autopilot
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill identifies build, test, and lint commands by reading project files such as Makefile, package.json, Cargo.toml, and pyproject.toml. The resulting commands are executed via omh_gather_evidence during the QA and Validation phases. This creates an attack surface where a malicious project configuration could lead to unauthorized command execution.\n
- Ingestion points: Project configuration files (Makefile, package.json, etc.) in the working directory.\n
- Boundary markers: None present to distinguish between trusted and untrusted project commands.\n
- Capability inventory: omh_gather_evidence performs shell command execution; delegate_task allows subagent operations.\n
- Sanitization: No explicit validation or filtering of the commands extracted from project files.\n- [COMMAND_EXECUTION]: The skill is designed to automatically execute project-specific build and test scripts. While necessary for its function as an autonomous developer, this grants the agent the ability to run arbitrary shell commands defined within the codebase it is modifying.
Audit Metadata