council
Pass
Audited by Gen Agent Trust Hub on Oct 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill's core functionality involves analyzing user-provided ideas and proposals through simulated debate. This ingestion of untrusted data creates a potential surface where malicious instructions could be embedded to influence the agent's behavior.\n
- Ingestion points: User-provided 'ideas', 'decisions', 'constraints', and 'stakes' are identified in the 'Frame and select' section of SKILL.md as the primary input for analysis.\n
- Boundary markers: The instructions do not define specific delimiters or isolation techniques (such as XML tags or unique markers) to separate the user-provided proposal from the skill's operational instructions.\n
- Capability inventory: The skill body includes a prompt-level guardrail ('Do not modify project files or start implementation unless separately requested'); however, the underlying agent platform may provide capabilities like shell access and file editing that an injection could attempt to exploit.\n
- Sanitization: There are no explicit instructions for validating or sanitizing user input before it is interpreted by the simulated personas in the council.
Audit Metadata