business-brainstorm
Pass
Audited by Gen Agent Trust Hub on Oct 1, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection as it processes and interpolates data from various external and user-controlled sources.\n
- Ingestion points: The skill accepts user-supplied business ideas (Step 1), reads local user configuration files like
portfolio.local.md(Step 2), accesses historical project files (project_*.md) from Memory (Step 2), and incorporates output from thedeep-researchtool (Step 4).\n - Boundary markers: The skill instructions do not define clear boundary markers or system instructions to disregard embedded commands when these external data sources are processed.\n
- Capability inventory: The skill has the capability to execute tool calls (
deep-research,/domain) and write data to the local file system (Step 7 - archiving).\n
- Sanitization: No sanitization, escaping, or validation logic is described for the content ingested from external or user-controlled sources.
Audit Metadata