Perform text-to-image or image-to-text search
AI agents call cross_modal_search to retrieve information from Jina AI MCP Server without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
This tool retrieves and queries data across different modalities (text and images) to find matches, which is characteristic of Read operations. There are no side effects, no data modification, no destructive actions, and no financial implications. The operation is passive information retrieval, making it the lowest-risk category.
From the tool's definition Tool performs "text-to-image or image-to-text search" - a query operation that retrieves matching results without modifying, deleting, or executing external code.
Attacks that exploit this kind of access
Perform text-to-image or image-to-text search. It is categorised as a Read tool in the Jina AI MCP Server MCP Server, which means it retrieves data without modifying state.
Register the Jina AI MCP Server MCP server in PolicyLayer and add a rule for cross_modal_search: allow, deny, rate-limit, or require approval. Point your MCP client at the PolicyLayer proxy URL and the rule is enforced on every call, before it reaches Jina AI MCP Server. Nothing to install.
cross_modal_search is a Read tool with low risk. Read-only tools are generally safe to allow by default.
Yes. Add a rate_limit block to the cross_modal_search rule in your PolicyLayer policy. For example, setting max: 10 and window: 60 limits the tool to 10 calls per minute. Rate limits are tracked per agent session and reset automatically.
Set action: deny in the PolicyLayer policy for cross_modal_search. The AI agent will receive a policy violation error and cannot call the tool. You can also include a reason field to explain why the tool is blocked.
cross_modal_search is provided by the Jina AI MCP Server MCP server (sheshiyer/jina-ai-mcp-multimodal-search). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Every MCP server has a record like this.
Type a name, get the same breakdown: verified identity, auth posture, risk grade, capabilities, recommended policy.
Teams ship this data inside their own products. See what a licence covers →