AI agents call test_connection to retrieve information from MCP RAG without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
This is a non-destructive probe operation that queries the status of an external service without creating, modifying, or deleting any data. It has minimal blast radius if misused by an agent—worst case, it generates network traffic or reveals service availability information. It fits clearly into the Read category as a diagnostic/status check with no state changes.
From the tool's definition Tool performs a connection test to the embedding service, which is a read-only diagnostic operation with no side effects. The description explicitly states it 'test[s] the connection' rather than modifying, executing, or deleting anything.
Attacks that exploit this kind of access
Test the connection to the embedding service. It is categorised as a Read tool in the MCP RAG MCP Server, which means it retrieves data without modifying state.
Register the MCP RAG MCP server in PolicyLayer and add a rule for test_connection: 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 MCP RAG. Nothing to install.
test_connection 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 test_connection 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 test_connection. 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.
test_connection is provided by the MCP RAG MCP server (santis84/mcp-rag). 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 →