clarvia-aeo-check
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill references the installation of an external MCP server package using
npx --yes clarvia-mcp-server@<reviewed-version>. This pattern is used for tool integration, and the inclusion of a version placeholder encourages the use of verified releases. - [EXTERNAL_DOWNLOADS]: The documentation provides an optional CI/CD integration using the GitHub Action
clarvia-project/clarvia-action@<reviewed-full-commit-SHA>. The use of a commit SHA placeholder follows security best practices for supply chain integrity. - [INDIRECT_PROMPT_INJECTION]: The skill ingests external data (URLs and names of third-party tools) to generate evaluation scores. While this creates a surface for indirect prompt injection if the external tool metadata contains malicious instructions, the skill mitigates this by explicitly stating that scores are advisory and should not be used as a substitute for security audits or manual review.
- [DATA_EXFILTRATION]: The 'Limitations' section explicitly warns that scanning a tool or URL sends data to the provider (Clarvia). This transparency ensures users are aware of the data flow before interacting with private or internal endpoints.
Audit Metadata