Kind: Server-side

The Model Endpoint Call tool turns another inference endpoint into a callable tool. When the model invokes it, Infersec runs an inference pass against the target endpoint using the tool's system prompt, passes the tool's arguments as input, and returns the model's output to the calling model. This lets you compose specialised models (a vision locator, a JSON extractor, a translator) into a single agent loop.

No Conduit agent is required - the call is executed entirely within the Infersec API.

How it works

  1. The model calls the tool with a set of arguments matching the tool's input schema.
  2. Infersec sends a chat completion to the referenced inference endpoint, applying the systemPrompt.
  3. The endpoint's response (optionally validated against the outputSchema) is returned as the tool result.

Configuration

Field Required Description
description yes What the tool does; shown to the model
endpointId yes The target inference endpoint to call
inputSchema no JSON schema describing the arguments the tool accepts
outputSchema no JSON schema describing the expected response
systemPrompt no System prompt applied to the inner inference call

The endpointId must reference an inference endpoint in the same account. Use a descriptive description and a tight systemPrompt - the inner model only sees those, not the outer conversation.

Connecting

Server-side tools need no Conduit connection. Create the tool in the console, add it to a tool service, and attach that tool service to an inference endpoint.