omc-reference

Pass

Audited by Gen Agent Trust Hub on Mar 24, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill documentation references tools with high-privilege capabilities, including python_repl for arbitrary code execution and ast_grep_replace for automated codebase modifications.
  • [PROMPT_INJECTION]: The skill defines a large attack surface for indirect prompt injection (Category 8) due to the integration of data-ingestion tools and autonomous execution workflows.
  • Ingestion points: The framework uses tools like state_read, project_memory_read, lsp_hover, and notepad_read to pull data into the agent context (SKILL.md).
  • Boundary markers: The skill does not define specific delimiters or 'ignore' instructions for data processed by these tools.
  • Capability inventory: Referenced tools include python_repl, ast_grep_replace, and autonomous loops such as autopilot and ralph (SKILL.md).
  • Sanitization: No sanitization or input validation mechanisms are described for the data ingested from the tools.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 24, 2026, 07:43 AM
Security Audit — agent-trust-hub — omc-reference