council
Pass
Audited by Gen Agent Trust Hub on Sep 24, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted project context and user-provided questions which can lead to indirect prompt injection.
- Ingestion points: Files in the workspace such as CLAUDE.md, documents in the memory/ folder, and the user's input decision string in SKILL.md.
- Boundary markers: Uses horizontal rules (---) and bold headers for response labeling, but lacks explicit guardrail instructions telling sub-agents to ignore instructions found within the processed text.
- Capability inventory: Performs file discovery (Glob) and reading within the current directory, and writes result transcripts to the local filesystem.
- Sanitization: External content is interpolated directly into agent prompts without validation or escaping.
Audit Metadata