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

# Connect OpenAI Codex to the Bland MCP server

> Connect OpenAI Codex to your Bland account with a config.toml entry, or install the Bland plugin.

The Bland MCP server is a hosted, streamable-HTTP server. Every client needs the same two values:

* **URL:** `https://api.bland.ai/v1/mcp`
* **Header:** `Authorization: Bearer YOUR_BLAND_API_KEY`

Get your key from the [dashboard](https://app.bland.ai/dashboard/settings). Setting up a different tool? See all [MCP clients](/integrations/mcp/clients).

## Setup

Prefer the [Bland plugin](/integrations/mcp/norm), which installs into Codex from the same marketplace. For the server alone, add to `~/.codex/config.toml`:

```toml theme={null}
[mcp_servers.bland]
url = "https://api.bland.ai/v1/mcp"
bearer_token_env_var = "BLAND_API_KEY"
```

Set `BLAND_API_KEY` in the environment that launches Codex, then restart Codex. For the CLI, export it in the same terminal before running `codex`:

```bash theme={null}
export BLAND_API_KEY="org_..."
codex
```

Replace `org_...` with your key. For the desktop app or IDE extension, the app's process must receive the variable too. Editing a shell profile alone may not make it available to an app launched from your desktop; use the Bland plugin if you prefer its credential setup.

If your Codex version has trouble with HTTP servers, use the stdio bridge on the [Any client](/integrations/mcp/clients/other) page.

## Next steps

<CardGroup cols={2}>
  <Card title="Tool reference" icon="wrench" href="/integrations/mcp/tools">
    Everything the MCP server can do.
  </Card>

  <Card title="Bland MCP Server" icon="circle-info" href="/integrations/mcp/overview">
    Auth, rate limits, and how the server works.
  </Card>
</CardGroup>

Docs for agents: [llms.txt](/llms.txt)