MCP with VS Code

This guide walks you through installing and configuring the raia MCP (Model Context Protocol) Server in Visual Studio Code. Once configured, VS Code can connect to raia's MCP endpoint and expose raia tools directly inside your editor.

• Visual Studio Code installed (v1.103+ recommended).
• A valid raia MCP endpoint URL.
• A valid raia API key.
• Internet access to reach https://api.raia2.com/mcp.

1. 01

   Open Extensions, search for "Model Context Protocol".

2. 02

   Install an MCP-compatible extension such as Azure MCP Server (Microsoft) or another extension that supports HTTP servers.

3. 03

   If prompted about Restricted Mode, click Trust to enable full functionality.

1. 01

   Open the Command Palette (Cmd+Shift+P on Mac, Ctrl+Shift+P on Windows).

2. 02

   Type "MCP: Add Server" and select it.

3. 03

   Choose server type "HTTP (HTTP or Server-Sent Events)".

4. 04

   When prompted, enter https://api.raia2.com/mcp and press Enter.

The raia MCP server requires an API key. In the Command Palette, run MCP: Open User Configuration to open your mcp.json file, then update it to look like this:

{
  "servers": {
    "raia-mcp": {
      "type": "http",
      "url": "https://api.raia2.com/mcp",
      "headers": {
        "Authorization": "YOUR_API_KEY_OF_AGENT_FROM_RAIA_MCP_SKILL"
      }
    },
    "NAME OF AGENT": {
      "url": "https://api.raia2.com/mcp",
      "type": "http"
    }
  },
  "inputs": []
}

1. 01

   After saving, open the Command Palette and run "MCP: List Servers".

2. 02

   Select raia-mcp and click Start Server.

3. 03

   Go to View → Output and select "MCP: raia-mcp" to view logs.

You should see logs similar to:

Starting server raia-mcp
Connection state: Running
Discovered 1 tools

In VS Code Settings, search for chat.mcp.autostart and set it to newAndOutdated. This prevents having to manually restart the MCP server.

If you see an error similar to:

Connection state: Error Error sending message to https://api.raia2.com/mcp: TypeError: Failed to fetch

This is typically caused by CORS in cloud-based VS Code (VS Code Web, Codespaces). Paste the instructions below into the AI Chat in VS Code to walk through the fix.

Step 1 — Locate Configuration Files

In cloud VS Code, settings are typically stored in the cloud profile. Find or create:

~/.config/Code/User/mcp.json
# or, for VS Code Web/Codespaces:
.vscode/mcp.json

Open the integrated terminal (Ctrl+~) and run:

ls ~/.config/Code/User/
ls .vscode/

Step 2 — Create or Update mcp.json

Create .vscode/mcp.json and paste:

{
  "servers": {
    "raia-mcp": {
      "type": "http",
      "url": "https://api.raia2.com/mcp",
      "headers": {
        "Authorization": "YOUR_RAIA_MCP_KEY"
      }
    }
  },
  "inputs": [
    "Use the RAIA MCP for researching code",
    "Call raia-mcp when the user asks about how the legacy code works"
  ]
}

Or via terminal:

mkdir -p ~/.config/Code/User
cat > ~/.config/Code/User/mcp.json << 'EOF'
{
  "servers": {
    "raia-mcp": {
      "type": "http",
      "url": "https://api.raia2.com/mcp",
      "headers": {
        "Authorization": "YOUR_RAIA_MCP_KEY"
      }
    }
  },
  "inputs": []
}
EOF

Step 3 — Enable MCP in Settings

Open Settings (Cmd/Ctrl+,) and enable chat.mcp.autostart and chat.mcp.enabled. Or edit settings.json:

{
  "chat.mcp.autostart": true,
  "chat.mcp.enabled": true
}

Step 4 — Verify the Configuration

cat ~/.config/Code/User/mcp.json
# or
cat .vscode/mcp.json

Test the MCP server is reachable:

curl -s -X POST "https://api.raia2.com/mcp" \
  -H "Authorization: YOUR_RAIA_MCP_KEY" \
  -H "Content-Type: application/json" \
  -H "Accept: application/json, text/event-stream" \
  -d '{"jsonrpc": "2.0", "method": "tools/list", "id": 1}' | head -20

Expected: a response starting with event: message or data: (not an error).

Step 5 — Reload VS Code

Press Cmd/Ctrl+Shift+P and run Developer: Reload Window, or close and reopen cloud VS Code in your browser.

Step 6 — Test the Connection

Open VS Code Chat (Cmd/Ctrl+Shift+I) and send a test message like:

Query RAIA MCP: What is the Quantum system?

If the built-in MCP client isn't working, you can manually query using Python:

import requests

url = "https://api.raia2.com/mcp"
headers = {
    "Authorization": "YOUR_RAIA_MCP_KEY",
    "Content-Type": "application/json",
    "Accept": "application/json, text/event-stream",
}
payload = {
    "jsonrpc": "2.0",
    "method": "tools/call",
    "params": {
        "name": "get-any-info-about-Code_Agent_-_Quantum",
        "arguments": {"prompt": "What is Quantum?"},
    },
    "id": 1,
}

response = requests.post(url, json=payload, headers=headers, stream=True)
for line in response.iter_lines():
    if line:
        print(line.decode("utf-8"))