Select an Agent Tool with Jev

The decision boundary

This pattern chooses a tool category from an allowlist. It does not execute the tool or grant permissions. Your application must still validate arguments, enforce authorization, and restrict the resources each tool can access.

Try your own context

Edit the example below. Selecting a preset does not use an allowance; only a successful run does.

What to expect

For a request about current Tokyo weather, Weather API is a plausible tool selection. This is an illustrative expectation, not a measured prediction. Run the example to see the real result.

Use it in your application

{
  "playground": "https://aijev.net",
  "language": "JavaScript",
  "request": {
    "model": "typesafe/jev-1.13",
    "state": "A user asks: What is the weather in Tokyo right now?",
    "questions": {
      "decision": {
        "type": "choice",
        "instructions": "Which tool should the agent use?",
        "criteria": {
          "Weather API": "Weather API",
          "Web Search": "Web Search",
          "Calculator": "Calculator",
          "No tool": "No tool"
        }
      }
    },
    "provider": {
      "allow_fallbacks": false
    }
  },
  "note": "AIJev Playground request shape only. For application integration, visit https://docs.typesafe.ai."
}

When this pattern is useful

Use this when the available tools have clear, distinct capabilities. Keep the choice set short and describe the difference between live lookup, general search, calculation, and no-tool responses.

When a human should review

Require review before irreversible or consequential actions such as publishing, purchasing, or deleting. Treat text in the state as untrusted input, and never allow a model’s tool choice to override your application’s permissions.

Frequently asked questions

Next: Support ticket routing · API integration guide · All examples.