---
title: Use Orchestor's MCP server
description: Connect Claude Code, Codex, Cursor, and other AI tools to Orchestor
  to explore brands, AI answers, and visibility reports.
canonical_url: https://orchestor.io/docs/en/mcp
markdown_url: https://orchestor.io/docs/en/mcp.md
contentType: how-to
---

# Use Orchestor's MCP server

Connect your AI tools to Orchestor using the [Model Context Protocol (MCP)](https://modelcontextprotocol.io/). Explore brands and AI answers, compare visibility, and investigate citation sources through conversations with your agent.

## Quick setup

Add this server URL to your MCP client:

```text
https://mcp.orchestor.io/mcp
```

Then start authentication from your client, sign in to Orchestor, and authorize access to the workspace you want to use.

[Set up your client](#set-up-your-client)

## What is Orchestor MCP?

Orchestor MCP is Orchestor's official MCP server. It gives AI tools remote access to your workspace without requiring you to run a server. It uses Streamable HTTP for the connection and OAuth for authentication.

Ask your agent to:

- Find brands, topics, and monitoring prompts.
- Inspect saved AI answers and their cited sources.
- Compare visibility, citation, and sentiment reports.
- Record an improvement as an Issue when you request it.

## Available tools

Orchestor MCP provides tools for reading brands, answers, and reports, as well as discovering additional read tools by purpose. The [tools reference](https://orchestor.io/docs/en/mcp/tools.md) lists tools by category with their inputs and sample requests.

Available operations depend on the server configuration and your connection's permissions. For example, `prompts_create` previews a prompt, while `create_issue` creates an Issue. A prompt preview does not save the prompt.

## Connect to Orchestor MCP

You need an Orchestor account with access to the workspace. Your client must support remote MCP and OAuth. You do not need to configure an API key for this connection.

Setup guides are available for:

- [Claude Code](#claude-code)
- [Codex](#codex)
- [Cursor](#cursor)
- [OpenCode](https://orchestor.io/docs/en/agent-setup/opencode.md)

## Set up your client

### Claude Code

Add the MCP server from your terminal:

```bash
claude mcp add --transport http orchestor https://mcp.orchestor.io/mcp
```

Start Claude Code, open `/mcp`, and authenticate `orchestor`. Confirm the workspace and permissions in your browser, then authorize access.

[Claude Code setup guide](https://orchestor.io/docs/en/agent-setup/claude-code.md)

### Codex

Add the server to `~/.codex/config.toml`:

```toml
[mcp_servers.orchestor]
url = "https://mcp.orchestor.io/mcp"
```

Authenticate and check the server registration:

```bash
codex mcp login orchestor
codex mcp list
```

[Codex setup guide](https://orchestor.io/docs/en/agent-setup/codex.md)

### Cursor

Add the server to your project's `.cursor/mcp.json`. If the file already contains servers, add `orchestor` to the existing `mcpServers` object.

```json
{
  "mcpServers": {
    "orchestor": {
      "url": "https://mcp.orchestor.io/mcp"
    }
  }
}
```

Start authentication for `orchestor` from Cursor's MCP settings and authorize access in your browser.

[Cursor setup guide](https://orchestor.io/docs/en/agent-setup/cursor.md)

## Try the connection

Ask your connected agent:

> List the brands in my Orchestor workspace. Include each brand's ID and name.

A successful brand-list response completes your first read. An empty list is expected if you have not added brands. Next, specify a brand and date range:

> Review this brand's visibility and citation sources over the last 30 days. Include the supporting reports and AI answers. Separate observations from recommendations.

See [use cases](https://orchestor.io/docs/en/mcp/use-cases.md) for more sample requests.

## Manage access and connections

On the consent screen, check the client and workspace. Choose **Read-only** for analysis or **Read and write** for changes such as creating an Issue. Choose **This session** or **Until revoked** for the connection duration.

Access is limited to the workspace selected during authorization. To use another workspace, authorize a connection to that workspace.

Open [MCP connections](https://app.orchestor.io/settings/mcp) to inspect or disconnect a client. Select **Revoke** to prevent that connection from accessing the workspace.

## Troubleshooting

| Situation | What to do |
| --- | --- |
| The server is added but cannot read data | Start authentication from your client and authorize workspace access in your browser. |
| Authorization has expired | Authenticate again from the client. Check the connection in MCP connections. |
| The expected data is missing | Check the authorized workspace, target brand, and date range. |
| Reads work but creating an Issue fails | Check that the connection has Read and write access. |
| A tool is missing | Check its purpose in the [tools reference](https://orchestor.io/docs/en/mcp/tools.md) and ask the agent to inspect the available tools. |

## Related resources

- [MCP tools reference](https://orchestor.io/docs/en/mcp/tools.md)
- [MCP use cases](https://orchestor.io/docs/en/mcp/use-cases.md)
- [Orchestor Skills](https://orchestor.io/docs/en/agent-resources/skills.md)
- [Orchestor CLI](https://orchestor.io/docs/en/cli.md)

---

[Documentation index](https://orchestor.io/docs/llms.txt)