create-agent-with-sanity-context
Pass
Audited by Gen Agent Trust Hub on May 7, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is a developer-focused toolkit and reference implementation provided by Sanity.io for building context-aware AI agents.
- [DATA_EXPOSURE]: The skill includes functionality to capture the user's current page content as markdown and take screenshots using 'html2canvas-pro'. This data is sent to the configured LLM provider (e.g., Anthropic) to provide the agent with visual and structural context. This behavior is clearly documented as the primary purpose of the 'Agent Context' feature.
- [DATA_EXPOSURE]: The 'Conversation Insights' feature uses a telemetry integration to save chat transcripts to the user's Sanity dataset. This is used for classification and analytics as described in the documentation.
- [EXTERNAL_DOWNLOADS]: The skill references and uses official packages from trusted organizations, including '@sanity/', '@ai-sdk/', and 'next'. These dependencies are standard for the described use case.
- [REMOTE_CODE_EXECUTION]: The skill implements the Model Context Protocol (MCP) via HTTP transport to communicate with Sanity's API. This enables the agent to execute structured queries (GROQ) and explore schema metadata to fulfill user requests.
Audit Metadata