Set up other AI apps (Cursor, Claude Code, VS Code and more)

Watch the steps

1. 1Platform Setup
2. 2Integrations
3. 3API Token
4. 4Create New API Token
5. 5MCP Read
6. 6MCP Write
7. 7Create Token
8. 8Connect AI Agent
9. 9Paste the config into your agent

Connect Cursor, Claude Code or VS Code to BCL (0:30)

Coding assistants and AI apps other than Claude and ChatGPT connect to BCL with the same server address, https://bcl.my/mcp. Some sign you in through the browser (OAuth), and some need a Personal Access Token from BCL. This page gives the setup for each app on the Connect AI Agent page.

For claude.ai, Claude Desktop and ChatGPT, see Connect Claude or ChatGPT to BCL.

Pick your app and sign-in method

BCL lists 15 apps under Platform Setup → Connect AI Agent, on the Quick start tab. Each one supports the methods below:

App	OAuth (browser sign-in)	Personal Access Token	Where the setting goes
Claude Desktop / claude.ai	Yes	No	Settings → Connectors → Add custom connector
ChatGPT	Yes	No	Settings → Apps & Connectors → Advanced → Developer mode → Create
Claude Code	Yes (recommended)	Yes	claude mcp add, or .mcp.json in the project root
Cursor	Yes (recommended)	Yes	~/.cursor/mcp.json (global) or .cursor/mcp.json (project)
VS Code (GitHub Copilot)	Yes	Yes	code --add-mcp, or .vscode/mcp.json
Codex	Yes	No	~/.codex/config.toml
Gemini CLI	Yes (recommended)	Yes	gemini mcp add, or ~/.gemini/settings.json
Windsurf	Yes (recommended)	Yes	~/.codeium/windsurf/mcp_config.json
Warp	Yes	Yes	Settings → Agents → MCP servers → + Add
Zed	Yes (recommended)	Yes	Agent Panel → Add Remote Server, or ~/.config/zed/settings.json
OpenCode	Yes	No	opencode.json or ~/.config/opencode/opencode.json
Cline	No	Yes	MCP Servers → Configure MCP Servers
JetBrains AI	No	Yes	Settings → Tools → AI Assistant → Model Context Protocol (MCP) → Add
Hermes Agent	No	Yes	~/.hermes/config.yaml
OpenClaw / Custom	No	Yes	openclaw mcp set, or any MCP SDK

With OAuth, the app opens BCL in your browser. Sign in, check the permissions and click Authorize, as described in Approve the connection. Access and revoking work the same as for Claude.

Create a token for MCP

Skip this section if your app uses OAuth. For a token, you must be an Admin:

1. Go to Platform Setup → Integrations and open the API Token tab.

2. Click Create New API Token.

   

3. Enter a Token Name that says where you will use it, such as “Cursor on office laptop”.

4. Under Permissions, tick MCP Read. Tick MCP Write too only if the app should make changes. Untick API Read and API Write unless the same token also calls the BCL API.

   

5. Choose an Expiration, for example 90 days.

6. Click Create Token, then copy the token from the list with the copy icon.

For more on tokens, including how to view or delete one, see Create an API token.

Caution

A token works for anyone who has it, with the permissions you ticked. Keep it out of shared files and screenshots. If a config file with a token goes into Git, delete that token in BCL and create a new one.

In every snippet below, replace <YOUR_PAT_HERE> with your token. Keep the word Bearer and the space before it.

Claude Code

Claude Code signs in through your browser. Add the server with one command:

claude mcp add --transport http bcl https://bcl.my/mcp

Then run /mcp in Claude Code, choose bcl and sign in with your BCL account.

For CI or a machine without a browser, use a token instead:

claude mcp add --transport http bcl https://bcl.my/mcp \
  --header "Authorization: Bearer <YOUR_PAT_HERE>"

To share the server with your team, commit a .mcp.json file in the project root. Do not put a token in it.

Cursor

Save this in ~/.cursor/mcp.json (all projects) or .cursor/mcp.json (one project), then turn the server on under Settings → MCP. Cursor shows Needs login next to the server: click it and sign in with your BCL account.

{
    "mcpServers": {
        "bcl": {
            "url": "https://bcl.my/mcp"
        }
    }
}

With a token instead:

{
    "mcpServers": {
        "bcl": {
            "url": "https://bcl.my/mcp",
            "headers": {
                "Authorization": "Bearer <YOUR_PAT_HERE>"
            }
        }
    }
}

VS Code

VS Code uses MCP in GitHub Copilot Agent Mode. Turn on Agent Mode in Copilot Chat, then run:

code --add-mcp '{"name":"bcl","type":"http","url":"https://bcl.my/mcp"}'

VS Code asks you to trust the server, then opens the browser to sign in. With a token, put this in .vscode/mcp.json. Note that VS Code uses servers, not mcpServers, and needs a type:

{
    "servers": {
        "bcl": {
            "type": "http",
            "url": "https://bcl.my/mcp",
            "headers": {
                "Authorization": "Bearer <YOUR_PAT_HERE>"
            }
        }
    }
}

Windsurf

Windsurf uses serverUrl instead of url. Save this in ~/.codeium/windsurf/mcp_config.json, then click Refresh in the Cascade MCP panel. With no headers, Windsurf opens the browser to sign in:

{
    "mcpServers": {
        "bcl": {
            "serverUrl": "https://bcl.my/mcp"
        }
    }
}

For a token, add "headers": { "Authorization": "Bearer <YOUR_PAT_HERE>" } next to serverUrl.

Zed

Zed connects to remote servers natively. The easiest way is Agent Panel → Settings → Add Remote Server. Or add this to ~/.config/zed/settings.json; with no headers, Zed asks you to sign in in the browser:

{
    "context_servers": {
        "bcl": {
            "url": "https://bcl.my/mcp"
        }
    }
}

For a token, add "headers": { "Authorization": "Bearer <YOUR_PAT_HERE>" } next to url.

Cline

Cline needs a token. The easiest way is to paste this into a Cline chat, with your token in place of the placeholder:

Add this MCP server to your Cline MCP settings: name "bcl", type "streamableHttp", url https://bcl.my/mcp, with an Authorization header of "Bearer <YOUR_PAT_HERE>". Then confirm the BCL tools are available.

Or add it by hand under MCP Servers → Configure MCP Servers. The type must be exactly streamableHttp:

{
    "mcpServers": {
        "bcl": {
            "type": "streamableHttp",
            "url": "https://bcl.my/mcp",
            "headers": {
                "Authorization": "Bearer <YOUR_PAT_HERE>"
            },
            "disabled": false,
            "autoApprove": []
        }
    }
}

JetBrains AI

IntelliJ IDEA, PhpStorm, WebStorm, PyCharm and the other JetBrains IDEs need a token. Go to Settings → Tools → AI Assistant → Model Context Protocol (MCP) → Add, choose As JSON and paste:

{
    "mcpServers": {
        "bcl": {
            "url": "https://bcl.my/mcp",
            "headers": {
                "Authorization": "Bearer <YOUR_PAT_HERE>"
            }
        }
    }
}

If the IDE does not pick up the header, use the mcp-remote bridge instead. It needs Node.js:

{
    "mcpServers": {
        "bcl": {
            "command": "npx",
            "args": [
                "-y",
                "mcp-remote",
                "https://bcl.my/mcp",
                "--header",
                "Authorization: Bearer <YOUR_PAT_HERE>"
            ]
        }
    }
}

Codex, Gemini CLI, OpenCode and Warp

These apps sign in through the browser.

Codex (CLI, desktop app and IDE extension share one file): add this to ~/.codex/config.toml, then run codex mcp login bcl.

[mcp_servers.bcl]
url = "https://bcl.my/mcp"
oauth_resource = "https://bcl.my/mcp"

Gemini CLI: run the command below. On first use it opens the browser to sign in.

gemini mcp add --transport http --scope user bcl https://bcl.my/mcp

With a token: gemini mcp add --transport http --scope user --header "Authorization: Bearer <YOUR_PAT_HERE>" bcl https://bcl.my/mcp. In ~/.gemini/settings.json, the key is httpUrl, not url.

OpenCode: add this to opencode.json, then run opencode mcp auth bcl.

{
    "$schema": "https://opencode.ai/config.json",
    "mcp": {
        "bcl": {
            "type": "remote",
            "url": "https://bcl.my/mcp"
        }
    }
}

Warp: under Settings → Agents → MCP servers → + Add, choose Streamable HTTP or SSE Server (URL) and paste:

{
    "mcpServers": {
        "bcl": {
            "url": "https://bcl.my/mcp"
        }
    }
}

Hermes Agent, OpenClaw and your own agent

These use a token.

Hermes Agent: add this to ~/.hermes/config.yaml, then restart the agent.

mcp_servers:
  bcl:
    url: "https://bcl.my/mcp"
    headers:
      Authorization: "Bearer <YOUR_PAT_HERE>"

OpenClaw: it defaults to SSE, so set the transport:

openclaw mcp set bcl '{"url":"https://bcl.my/mcp","headers":{"Authorization":"Bearer <YOUR_PAT_HERE>"},"transport":"streamable-http"}'

Your own agent: point your MCP SDK’s Streamable HTTP client at https://bcl.my/mcp with the Authorization: Bearer header. To test the connection, list the tools:

curl -X POST https://bcl.my/mcp \
  -H "Authorization: Bearer <YOUR_PAT_HERE>" \
  -H "Content-Type: application/json" \
  -d '{"jsonrpc":"2.0","method":"tools/list","id":1}'

A token with MCP Read and MCP Write gets back 62 tools. BCL answers every request as plain JSON over POST; it does not open an event stream. For the tool list, limits and errors, see MCP tools reference.

Check that it works

After you save the settings, restart the app and ask it:

What tools do you have? Summarise what I can ask you to do.

It should list the BCL tools. Each request then shows in Tools → MCP Audit Log, with pat or oauth in the Auth column. See Check what your AI agent did.

Tips

A few habits make tokens and connections easier to manage:

• One token per app and computer. If a laptop is lost, you delete only that token and the other apps keep working.
• Prefer OAuth where you can. There is no token to copy, and you can revoke the app under Connected Apps.
• Set an expiry on tokens. A token with Never keeps working until someone deletes it.

Common issues

Which is better, OAuth or a Personal Access Token?

Use OAuth when your app supports it: you sign in with your BCL account in the browser and there is no token to copy or leak. Use a token for JetBrains AI, Cline, Hermes Agent, OpenClaw, scripts and machines without a browser.

My token works with the BCL API but the AI app says it is not allowed. Why?

MCP needs its own permissions. The token must have MCP Read (and MCP Write to make changes). API Read and API Write alone are refused.

The AI app connects but shows no BCL tools. What do I check?

Check the address is exactly https://bcl.my/mcp, that you replaced <YOUR_PAT_HERE> with your token, and that the token is not expired. Then restart the app and ask "What tools do you have?".

Related articles

• Connect Claude or ChatGPT to BCL
• Create an API token
• MCP tools reference
• Check what your AI agent did