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

# MCP Server

> Let AI agents read your crawl analytics and manage your Hado SEO configuration

> [!NOTE]
>
> The MCP server is currently in **beta**. Tools and response shapes are
> subject to change.

The Hado SEO MCP server lets AI agents — Claude, Claude Code, Cursor, Codex, and any other [Model Context Protocol](https://modelcontextprotocol.io) client — work with your account the same way the [REST API](https://docs.hadoseo.com/api-reference/introduction.md) does: read crawl analytics, export raw logs, and manage cache rules, blocked paths, and routing rules.

```
https://api.hadoseo.com/functions/v1/mcp
```

Because every MCP tool runs through the same API layer as the dashboard and REST endpoints, an agent can't do anything your API key couldn't do with plain HTTP — the same validation, plan limits, domain permissions, and rate limits apply.

## What agents can do

> [card] Analyze crawl activity
>
> Daily crawl rollups per bot, page-level crawl counts, AI referral traffic,
> and raw per-request crawl logs

> [card] Manage caching
>
> List, create, update, and delete per-path cache rules; purge and re-render
> pages on demand

> [card] Control crawl access
>
> Block paths from prerendering, with optional noindex/nofollow robots
> directives

> [card] Manage routing
>
> Create, update, and delete redirects and proxy rules on your live domain

See [MCP Tools](https://docs.hadoseo.com/api-reference/mcp/tools.md) for the full list of the 19 tools and their parameters, and [MCP Setup](https://docs.hadoseo.com/api-reference/mcp/setup.md) to connect a client.

## Definitions

| Term | Meaning |
| - | - |
| **MCP** | The Model Context Protocol — an open standard that lets AI applications call tools exposed by a server. The Hado SEO server uses the HTTP ("Streamable HTTP") transport. |
| **Tool** | One action an agent can take, such as `list_domains` or `create_cache_rule`. Read tools never change anything; write tools change your live domain within seconds. |
| **Scope** | What an API key is allowed to do: `read` (analytics and listing config) and/or `write` (changing config and purging). See [Key scopes](#key-scopes). |
| **Role** | Your relationship to a domain: `owner`, `manager` (can change config), or `viewer` (read-only). Shared domains appear in `list_domains` with your role. |
| **Window** | The date range for analytics tools: either a rolling `period` such as `"7d"` or `"30d"`, or an explicit `startDate`/`endDate` (`YYYY-MM-DD`, UTC). |
| **Cursor** | An opaque pagination token. When a response includes one, pass it back unchanged to fetch the next page. |

## Authentication

The server authenticates with the same [API keys](https://docs.hadoseo.com/api-reference/authentication.md) as the REST API, sent as a header when you connect the client:

```
Authorization: Bearer hado_sk_your_key_here
```

> [!NOTE]
>
> The MCP server is available on **every plan**. Plans differ by monthly call
> quota — see [Plan Limits](https://docs.hadoseo.com/api-reference/authentication.md#plan-limits).

Every tool call counts as **one API request** against your key's rate limits (60 requests/minute, plus your plan's monthly quota — from 300 calls/month on Starter up to 12,000 on Agency). Failed calls are not charged.

### Key scopes

API keys carry scopes that limit what connected agents can do:

| Scope | Allows |
| - | - |
| `read` | All analytics tools and the `list_*` config tools |
| `write` | Creating, updating, and deleting config, and `purge_cache` |

Keys have both scopes by default. **For agents, create a read-only key unless you specifically want the agent changing configuration** — MCP keys end up in agent config files, which are easy to leak.

## Safety notes

> [!WARNING]
>
> Write tools change your **live domain** within seconds. Routing rules in
> particular affect real visitor traffic, not just bots. Give agents a
> read-only key by default, and when you do grant `write`, review what the
> agent plans to change before approving tool calls.

Every write made through the MCP server is recorded in an audit log — the tool used, its arguments, the API key, and the outcome — so changes made by agents are always traceable. If something looks wrong, [contact support](mailto:support@hadoseo.com) and we can tell you exactly what changed and when.


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