create-agent-with-sanity-context
Pass
Audited by Gen Agent Trust Hub on Feb 24, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill requires installing several packages from official registries (NPM), including
@sanity/agent-context,@ai-sdk/mcp, and the Vercel AI SDK. These resources are from trusted technology providers and are necessary for the skill's integration features. - [COMMAND_EXECUTION]: Provides instructions for using the Sanity CLI for project configuration and an exploration tool to validate how the AI agent interprets structured content.
- [DATA_EXFILTRATION]: Includes utility functions to capture webpage context, such as Markdown summaries of page text and visual screenshots. This information is intended to be passed to the LLM to provide the agent with situational awareness, which is the primary purpose of the provided 'Agent Context' patterns.
- [PROMPT_INJECTION]: The skill demonstrates best practices for handling untrusted content by using clear XML-style delimiters (e.g., ) in system prompts to distinguish between agent instructions and retrieved data, reducing the risk of indirect prompt injection.
Audit Metadata