oma-brainstorm

Pass

Audited by Gen Agent Trust Hub on Oct 6, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill follows design-first best practices, including constraints such as 'one question at a time' and requiring explicit user approval for each design section before persistence. It focuses on documentation and reasoning rather than code execution.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user ideas and project context to generate architectural designs. Although this creates a surface for indirect instructions, it is inherent to the brainstorming function.
  • Ingestion points: User prompts and codebase context are ingested during the PREPARE and ACQUIRE phases as defined in SKILL.md.
  • Boundary markers: The skill does not implement specific delimiters or 'ignore' instructions to isolate ingested data from agent logic.
  • Capability inventory: The skill possesses the ability to read the codebase (READ) and write design artifacts to the 'docs/plans/designs/' directory (WRITE).
  • Sanitization: The instructions do not specify any sanitization, filtering, or escaping protocols for user-provided content.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 6, 2026, 04:52 PM