probe
Pass
Audited by Gen Agent Trust Hub on Sep 3, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill references several configuration and guidance documents hosted on the author's GitHub repository (raw.githubusercontent.com/csark0812/toolbox/). These links provide the agent with standardized schemas and research data.
- [INDIRECT_PROMPT_INJECTION]: The instructions explicitly define an authority boundary, requiring the agent to treat all external tool outputs and documents as untrusted evidence rather than instructions. This design pattern specifically addresses the risk of indirect prompt injection by limiting the agent's ability to be influenced by malicious content in processed data.
Audit Metadata