brainstorm
Pass
Audited by Gen Agent Trust Hub on Nov 10, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection by ingesting and processing data from external or untrusted sources.
- Ingestion points: The skill uses the
WebSearchtool to gather market and competitor context (Phase 0.2) and reads project-specific files fromdocs/specs/anddocs/project-referencefor domain context. - Boundary markers: The instructions do not specify any delimiters or warnings to the model to ignore embedded instructions within the fetched web content or project documents.
- Capability inventory: The agent has the capability to perform network operations (
WebSearch), interact with other agents (planner,docs-manager), and hand off tasks to other skills like$initiative,$work-item, and$planwhich may involve file writing or further external interactions. - Sanitization: No evidence of input validation, escaping, or filtering of content retrieved from the web or local files was found in the instructions.
Audit Metadata