brainstorming
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted input in the form of a 'framed problem' and interpolates it directly into prompt templates for four different LLM personas.
- Ingestion points: The
{framing}and{options}variables in the debate prompt templates ingest user-provided content into the agent context. - Boundary markers: No explicit delimiters or 'ignore embedded instructions' warnings are used to wrap the interpolated user content.
- Capability inventory: The skill uses
Task(internal model calls),Write(filesystem access), andmcp__great_cto_llm_router__ask_kimi(external model routing) to process and output the results. - Sanitization: There is no evidence of sanitization, escaping, or validation performed on the user-provided framing before it is sent to the LLM panelists.
Audit Metadata