Framework integrations

Vercel AI SDK

Wrap any Route Tools call as a tool — the model picks when to call it; we pick the provider:

import { generateText, tool, stepCountIs } from 'ai';
import { z } from 'zod';
import { RouteTools } from 'route-tools';

const rt = new RouteTools({ apiKey: process.env.ROUTE_TOOLS_API_KEY! });

const { text } = await generateText({
  model: 'anthropic/claude-fable-5',
  prompt: 'What changed in the latest Cloudflare Workers release?',
  stopWhen: stepCountIs(5),
  tools: {
    webSearch: tool({
      description: 'Search the web for current information.',
      inputSchema: z.object({ query: z.string() }),
      execute: async ({ query }) => (await rt.search({ query })).result,
    }),
    readPage: tool({
      description: 'Fetch a URL and return its content as markdown.',
      inputSchema: z.object({ url: z.string().url() }),
      execute: async ({ url }) => (await rt.scrape({ url })).result,
    }),
  },
});

LangChain (Python)

from langchain_core.tools import tool
from route_tools import RouteTools

rt = RouteTools(api_key=os.environ["ROUTE_TOOLS_API_KEY"])

@tool
def web_search(query: str) -> str:
    """Search the web for current information."""
    res = rt.search(query=query)
    return "\n".join(f"{r['title']} — {r['url']}\n{r['snippet']}" for r in res["result"]["results"])

@tool
def read_page(url: str) -> str:
    """Fetch a URL and return its content as markdown."""
    return rt.scrape(url=url)["result"]["content"]

agent = create_react_agent(model, tools=[web_search, read_page])

MCP clients (Claude, Cursor, anything)

Skip the glue code entirely — connect to the hosted MCP server and get all five tools:

{
  "mcpServers": {
    "route-tools": {
      "url": "https://api.route.tools/mcp",
      "headers": { "Authorization": "Bearer rt_live_..." }
    }
  }
}

See MCP setup for per-client instructions.

OpenAI-style function calling

Use the OpenAPI spec to generate function definitions, or hand-write them — the request schemas are small and stable. Execute the call with a plain fetch/requests POST to /v1/tools/{category}.