NYC

council-of-five

Pass

Audited by Gen Agent Trust Hub on Feb 15, 2026

Risk Level: LOWNO_CODE
Full Analysis
  • [NO_CODE] (INFO): The skill consists entirely of natural language instructions and prompt templates. No Python, JavaScript, or shell scripts are included.
  • [SAFE] (SAFE): No patterns of malicious behavior, such as data exfiltration, remote code execution, or credential theft, were identified.
  • [INDIRECT_PROMPT_INJECTION] (LOW): The skill processes external user content by interpolating it into subagent prompts. 1. Ingestion points: User input for problem descriptions and current solutions. 2. Boundary markers: None; placeholders are used without explicit delimiters. 3. Capability inventory: None; subagents only generate text and lack tool access. 4. Sanitization: None. The impact is limited to generating unexpected text within the user's conversation session.
Audit Metadata
Risk Level
LOW
Analyzed
Feb 15, 2026, 11:58 PM