> ## Documentation Index
> Fetch the complete documentation index at: https://docs.olostep.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Olostep + Composio

> Give agents built with Composio live web search, scraping, crawling, batches and answers through the Olostep toolkit.

The [Olostep toolkit on Composio](https://composio.dev/toolkits/olostep) gives your agents 16 Olostep tools, either through a hosted MCP server or directly in your agent code. [Olostep](https://www.olostep.com) is web data infrastructure for AI agents and enterprise AI workflows. Its APIs provide live web search, scraping and crawling to extract, clean and structure data from the web.

[View on Composio →](https://composio.dev/toolkits/olostep)

## Before you start

* **An Olostep account with an API key:** [get one free](https://www.olostep.com/dashboard/api-keys), no credit card required. Your first 500 credits are included.
* **A Composio account and API key:** sign up at [composio.dev](https://composio.dev).

Composio asks for your Olostep API key when you connect the toolkit and stores it encrypted.

## Use the tools over MCP

Composio creates an MCP server with the Olostep toolkit for each of your users. This example connects it to the Vercel AI SDK:

```bash theme={null}
npm install @composio/core ai @ai-sdk/mcp @ai-sdk/openai
```

```ts theme={null}
import { Composio } from "@composio/core";
import { createMCPClient } from "@ai-sdk/mcp";
import { openai } from "@ai-sdk/openai";
import { generateText, stepCountIs } from "ai";

const composio = new Composio();
const { mcp } = await composio.create("your-user-id", {
  toolkits: ["olostep"],
  mcp: true,
});

const client = await createMCPClient({
  transport: { type: "http", url: mcp.url, headers: mcp.headers },
});

const { text } = await generateText({
  model: openai("gpt-5.6-sol"),
  tools: await client.tools(),
  prompt: "Crawl competitor website for pricing changes",
  stopWhen: stepCountIs(10),
});

console.log(text);
await client.close();
```

## Use the tools directly

You can also pass the tools to your model provider without MCP. This example uses the OpenAI Responses API:

```bash theme={null}
npm install @composio/core @composio/openai openai
```

```ts theme={null}
import OpenAI from "openai";
import { Composio } from "@composio/core";
import { OpenAIResponsesProvider } from "@composio/openai";

const composio = new Composio({ provider: new OpenAIResponsesProvider() });
const client = new OpenAI();

const session = await composio.create("your-user-id", { toolkits: ["olostep"] });
const tools = await session.tools();

let response = await client.responses.create({
  model: "gpt-5.6-sol",
  tools,
  input: [{ role: "user", content: "Crawl competitor website for pricing changes" }],
});

while (response.output.some((o) => o.type === "function_call")) {
  const outputs = await composio.provider.handleToolCalls(session, response.output);
  response = await client.responses.create({
    model: "gpt-5.6-sol",
    tools,
    previous_response_id: response.id,
    input: outputs,
  });
}

console.log(response.output_text);
```

See [Composio's documentation](https://docs.composio.dev) for other frameworks and for Python.

## Tools

| Tool | What it does |
| - | - |
| `OLOSTEP_SEARCH_WEB` | Search the web and return links |
| `OLOSTEP_SCRAPE_PAGE` | Scrape one page and return its content, links and metadata |
| `OLOSTEP_ANSWER_QUESTION` | Research a question on the web and return an answer with sources |
| `OLOSTEP_MAP_WEBSITE` | List a website's URLs, optionally ranked by relevance |
| `OLOSTEP_START_BATCH` | Start a batch scrape of many URLs and return a batch ID |
| `OLOSTEP_LIST_BATCH_ITEMS` | Get one page of a batch's results |
| `OLOSTEP_START_CRAWL` | Start a crawl of a website and return a crawl ID |
| `OLOSTEP_LIST_CRAWL_PAGES` | Get one page of the URLs a crawl processed |
| `OLOSTEP_RETRIEVE_CONTENT` | Get a scraped page's content by its retrieve ID |
| `OLOSTEP_GET_OPERATION_RESULT` | Get a stored scrape, search, map, answer, crawl or batch by its ID |
| `OLOSTEP_CREATE_MONITOR` | Monitor a page or a web query on a schedule, with optional alerts and a webhook |
| `OLOSTEP_LIST_MONITORS` | List your monitors and their settings |
| `OLOSTEP_LIST_MONITOR_EVENTS` | Get a monitor's snapshots and change summaries |
| `OLOSTEP_UPDATE_MONITOR` | Change a monitor's schedule, notifications or webhook |
| `OLOSTEP_SET_MONITOR_STATE` | Pause or resume a monitor |
| `OLOSTEP_GET_CREDIT_BALANCE` | Check your remaining credits and plan |

Batches and crawls run in the background. Start one, then use `OLOSTEP_LIST_BATCH_ITEMS` or `OLOSTEP_LIST_CRAWL_PAGES` and `OLOSTEP_RETRIEVE_CONTENT` to read the pages. The Olostep API takes up to 100k URLs in a single batch, and you can scale to 1M requests with 10 parallel batches.

## Resources

> [card] [Olostep on Composio](https://composio.dev/toolkits/olostep)
>
> The Olostep toolkit page

> [card] [Composio docs](https://docs.composio.dev)
>
> Sessions, MCP and frameworks

> [card] [Olostep website](https://www.olostep.com)
>
> Plans and pricing

> [card] [Support](mailto:info@olostep.com)
>
> [info@olostep.com](mailto:info@olostep.com)


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.