patent-search
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is purely instructional and directs the agent to use predefined MCP tools for BigQuery and Google Patent searches. It explicitly discourages the use of standalone scripts or Python execution, reducing the attack surface.
- [SAFE]: No hardcoded credentials were found. While the documentation mentions the requirement for a
SERPAPI_API_KEY, it does not provide one, following standard configuration practices. - [INDIRECT_PROMPT_INJECTION]: The skill processes external data (patent titles, abstracts, and claims) retrieved from BigQuery and Google. This presents a potential surface for indirect prompt injection if malicious instructions were embedded in patent metadata; however, the skill lacks high-privilege capabilities (like file writing or shell execution) that could be leveraged by such an injection.
- [SAFE]: No obfuscation, persistence mechanisms, or unauthorized privilege escalation patterns were detected in the instructions or metadata.
Audit Metadata