competitive-intel-agent
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFENO_CODE
Full Analysis
- NO_CODE (SAFE): The skill files provide only textual instructions for the agent's behavior. No scripts, binaries, or configuration files capable of executing code are present.
- CREDENTIALS_UNSAFE (SAFE): No hardcoded credentials, API keys, or sensitive environment variables were found in any of the analyzed files.
- INDIRECT_PROMPT_INJECTION (SAFE): The agents are designed to process external competitive data, which is an inherent vulnerability surface. However, the risk is mitigated because the skill contains no functional capabilities. Evidence chain: 1. Ingestion points: Product features, pricing, and market data provided in user queries. 2. Boundary markers: Absent in the system prompts. 3. Capability inventory: No tool access, network operations, or file-system interactions are defined. 4. Sanitization: Not applicable as the agent produces text-only structured analysis.
Audit Metadata