# ElevenAgents

> Give ElevenLabs voice and chat agents real-time web access with Firecrawl

Give your [ElevenAgents](https://elevenlabs.io/agents) voice and chat agents real-time web access using Firecrawl. This guide covers two integration paths:

1. **MCP server** — connect the hosted Firecrawl MCP server for zero-code setup.
2. **Server webhook tool** — point a custom tool at Firecrawl's REST API for full control over requests.

## Prerequisites

- An [ElevenLabs](https://elevenlabs.io) account with access to ElevenAgents
- A Firecrawl API key from your [firecrawl.dev dashboard](https://firecrawl.dev)

## Option 1: Firecrawl MCP Server

The fastest way to give an agent web access. ElevenAgents supports remote MCP servers, and Firecrawl provides a hosted MCP endpoint.

### Add the MCP server

1. Open the [Integrations page](https://elevenlabs.io/app/agents/integrations) in ElevenLabs and click **+ Add integration**.
2. Select **Custom MCP Server** from the integration library.
3. Fill in the following fields:

| Field | Value |
| --- | --- |
| **Name** | Firecrawl |
| **Description** | Search, scrape, and parse web content with Firecrawl. |
| **Server type** | Streamable HTTP |
| **Server URL** | `https://mcp.firecrawl.dev/v2/mcp` |

Keyless MCP exposes exactly Search, Scrape, and Parse, with shared limits. [Sign in](/mcp-server/oauth) or [add an API key](/mcp-server/keyless#add-an-api-key) for the broader tool surface. Keep API keys in an environment variable or secret store, never in a URL.

<Note>
You must select **Streamable HTTP** as the server type. The default SSE option does not work with the Firecrawl MCP endpoint.
</Note>

4. Under **Tool Approval Mode**, choose an approval level:
   - **No Approval** — the agent uses tools freely. Fine for read-only scraping.
   - **Fine-Grained Tool Approval** — lets you pre-select which tools can run automatically and which require approval.
   - **Always Ask** (default) — the agent requests permission before every tool call.

5. Check **I trust this server**, then click **Add Server**.

ElevenLabs will connect to the keyless server and list Search, Scrape, and Parse.

### Attach it to an agent

1. Create or open an agent in the [ElevenAgents dashboard](https://elevenlabs.io/app/agents/agents).
2. Go to the **Tools** tab, then select the **MCP** sub-tab.
3. Click **Add server** and select the **Firecrawl** integration from the dropdown.

### Update the system prompt

In the **Agent** tab, add instructions to the **System prompt** so the agent knows when to use Firecrawl. For example:

```text
You are a helpful research assistant. When the user asks about a website,
a company, or any topic that requires up-to-date information, use the
Firecrawl tools to search the web or scrape the relevant page, then
summarize the results.
```

### Test it

Click **Preview** in the top navigation bar. You can test using the text chat input or by starting a voice call. Try a prompt like:

> "What does firecrawl.dev do? Go to the site and summarize it for me."

The agent will call the Firecrawl MCP `scrape` tool, receive the page markdown, and respond with a summary.

---

## Option 2: Server Webhook Tool

Use this approach when you need precise control over request parameters (formats, headers, timeouts, etc.) or want to call a specific Firecrawl endpoint without exposing the full MCP tool set.

### Scrape tool

Create a tool that scrapes a single URL and returns its content as markdown.

1. Open your agent and go to the **Tools** tab.
2. Click **Add tool** and select **Webhook**.
3. Configure the tool:

| Field | Value |
| --- | --- |
| **Name** | scrape_website |
| **Description** | Scrape content from a URL and return it as clean markdown. |
| **Method** | POST |
| **URL** | `https://api.firecrawl.dev/v2/scrape` |

<Note>
The **Method** field defaults to GET — make sure to change it to **POST**.
</Note>

4. Scroll to the **Headers** section and click **Add header** for authentication:

| Header | Value |
| --- | --- |
| `Authorization` | `Bearer YOUR_FIRECRAWL_API_KEY` |

Alternatively, if you have workspace auth connections configured, you can use the **Authentication** dropdown instead.

5. Add a **body parameter**:

| Parameter | Type | Description | Required |
| --- | --- | --- | --- |
| `url` | string | The URL to scrape | Yes |

6. Click **Add tool**.

The Firecrawl API returns the page content as markdown by default. The agent receives the JSON response and can use the `markdown` field to answer questions.

### Search tool

Create a tool that searches the web and returns results with scraped content.

1. Click **Add tool** → **Webhook** again and configure:

| Field | Value |
| --- | --- |
| **Name** | search_web |
| **Description** | Search the web for a query and return relevant results with page content. |
| **Method** | POST |
| **URL** | `https://api.firecrawl.dev/v2/search` |

2. Add the same `Authorization` header as above.

3. Add **body parameters**:

| Parameter | Type | Description | Required |
| --- | --- | --- | --- |
| `query` | string | The search query | Yes |
| `limit` | number | Maximum number of results to return (default 5) | No |

4. Click **Add tool**.

### Update the system prompt

In the **Agent** tab, update the **System prompt**:

```text
You are a knowledgeable assistant with access to web tools.

- Use `scrape_website` when the user gives you a specific URL to read.
- Use `search_web` when the user asks a general question that requires
  finding information online.

Always summarize the information concisely and cite the source URL.
```

### Test it

Click **Preview** and try asking:

> "Search for the latest Next.js features and give me a summary."

The agent will call `search_web`, receive results from Firecrawl, and respond with a summary of the findings.

---

## Tips

- **Model selection** — For reliable tool calling, use a high-intelligence model such as GPT-4o, Claude Sonnet 4.5 or later, or Gemini 2.5 Flash. Smaller models may struggle to extract the correct parameters.
- **Keep prompts specific** — Tell the agent exactly when to use each tool. Vague instructions lead to missed or incorrect tool calls.
- **Limit response size** — For voice agents, long scraped pages can overwhelm the LLM context. Use `onlyMainContent: true` in scrape options (or instruct the agent to summarize aggressively) to keep responses concise.
- **Tool call sounds** — In the webhook or MCP tool settings, you can configure a **Tool call sound** to play ambient audio while a tool runs. This signals to the user that the agent is working.

## Resources

- [ElevenAgents documentation](https://elevenlabs.io/docs/eleven-agents/overview)
- [ElevenAgents tools overview](https://elevenlabs.io/docs/eleven-agents/customization/tools)
- [ElevenAgents MCP integration](https://elevenlabs.io/docs/eleven-agents/customization/tools/mcp)
- [Firecrawl API reference](https://docs.firecrawl.dev/api-reference/v2-introduction)
- [Firecrawl MCP server](https://docs.firecrawl.dev/mcp-server)