Skills Quick Start

Add a built-in skill to a chat completion in 30 seconds

Pass a skills array alongside your normal chat completion request. TheRouter injects the skill's tools, executes them server-side when the model calls them, and returns a final response with annotations.

Send a request

curl

curl https://api.therouter.ai/v1/chat/completions \
  -H "Authorization: Bearer $THEROUTER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "anthropic/claude-sonnet-4.5",
    "messages": [{ "role": "user", "content": "Latest advances in quantum computing." }],
    "skills": [{ "id": "web" }]
  }'

Response format

The response is a standard chat completion. When the web skill finds sources it adds url_citation annotations to the assistant message.

response.json

{
  "choices": [{
    "message": {
      "role": "assistant",
      "content": "Recent breakthroughs in quantum computing include ...",
      "annotations": [
        {
          "type": "url_citation",
          "url_citation": {
            "url": "https://example.com/quantum-2026",
            "title": "Quantum Computing Advances in 2026"
          }
        }
      ]
    }
  }]
}

TypeScript SDK

TypeScript

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://api.therouter.ai/v1",
  apiKey: process.env.THEROUTER_API_KEY,
});

const completion = await client.chat.completions.create({
  model: "anthropic/claude-sonnet-4.5",
  messages: [{ role: "user", content: "Latest advances in quantum computing." }],
  skills: [{ id: "web" }],
} as never);

const message = completion.choices[0]?.message;
console.log(message?.content);
console.log(message?.annotations);

Structured JSON outputs?

Try the response-healing skill when you use response_format: json_schema. It auto-repairs malformed JSON before your parser sees it.