brainstorm
Pass
Audited by Gen Agent Trust Hub on Jun 19, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill maintains an attack surface for indirect prompt injection by ingesting data from untrusted sources to inform its brainstorming process.
- Ingestion points: User arguments ($ARGUMENTS) in SKILL.md, local codebase files via Read, Grep, and Glob tools, and external web content via WebFetch.
- Boundary markers: No specific delimiters or instructions are provided to the agent to distinguish between its primary instructions and the ingested content.
- Capability inventory: Tool access is limited to read-only file operations, web searching, and restricted bash commands (git only). It lacks file-writing or arbitrary command execution capabilities.
- Sanitization: No explicit filtering or validation is performed on the ingested data. However, the risk is mitigated as the skill's primary function is informational and conversational, and its toolset is appropriately restricted.
Audit Metadata