idea-generator
Pass
Audited by Gen Agent Trust Hub on Aug 2, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill uses
WebSearchandWebFetchtools to gather market research and user complaints from external sources like Reddit. This introduces a risk of indirect prompt injection, as malicious instructions embedded in web content could potentially influence the agent's behavior during the brainstorming process. - Ingestion points:
WebSearchandWebFetchoperations described in Step 3 (Problem-First Lens) and Step 5 (Market Gap Lens) ofSKILL.md. - Boundary markers: No explicit delimiters or instructions to ignore embedded commands are present for the external data.
- Capability inventory: Access to
Write(to saveidea-shortlist.json),WebSearch,WebFetch, andAskUserQuestiontools. - Sanitization: The instructions do not define any validation, filtering, or escaping logic for the retrieved web content before it is processed by the agent.
Audit Metadata