Sigrid MCP Integrations

Sigrid MCP integrations let AI coding tools use Sigrid’s analysis while you work — catching security and quality issues as they’re introduced, and driving down technical debt through guided refactoring.

• Sigrid Guardrails MCP: Leverage Sigrid’s code analysis to safeguard AI Coding Assistants from introducing security and other quality issues
• Sigrid Modernization Recipes MCP: Use data from Sigrid to let AI Coding Agents perform large-scale modernization tasks

Start with Guardrails for automatic quality checks on new code. Add Recipes when you’re ready to tackle existing technical debt.

Installation

Claude Code plugin (recommended)

The Sigrid AI Toolkit is a Claude Code plugin that automatically configures both the Sigrid MCP server and the associated skills. This is the easiest way to get started.

/plugin marketplace add Software-Improvement-Group/sigrid-ai-toolkit
/plugin install sigrid@sigrid-ai-toolkit

The installer will prompt for your Sigrid API token and store it securely in the system keychain.

• Step 1: Obtain a Sigrid Token — see the Sigrid docs on authentication tokens
• Step 2: Run the two commands above in Claude Code
• Step 3: Follow the installer prompts

Manual configuration (other IDEs)

For IDEs other than Claude Code, configure the MCP server manually:

• Step 1: Obtain a Sigrid Token — see the Sigrid docs on authentication tokens
• Step 2: Configure the MCP tool in your IDE or AI Coding Assistant using the instructions below

Supported IDEs

Connection types explained

Direct HTTP

• Simplest configuration
• IDE connects directly to MCP server
• Used by: Cursor

HTTP via GitHub Copilot Extension

• Requires GitHub Copilot plugin
• Configuration through extension UI
• Used by: VSCode, IntelliJ family

Proxy (mcp-remote)

• Uses npx mcp-remote (install globally first if needed: npm install -g mcp-remote)
• Required when direct HTTP not supported
• Used by: Windsurf, VSCode

Cursor / GitHub Copilot Plugin

• Open IDE
• Click MCP & Integrations panel (left sidebar)
• Paste configuration:

{
  "mcpServers": {
    "SigridCode": {
      "url": "https://sigrid-says.com/mcp",
      "headers": {
        "Authorization": "Bearer <your_sigrid_token>"
      }
    }
  }
}

VSCode

• Install Node (needed for the npx package)
• Install GitHub Copilot extension
• Connect your GitHub account
• In main screen of VSCode, click the settings icon on bottom left, select “Profiles”
• Click MCP Servers. If prompted to create a new file, say “Yes”

Add:

{
  "servers": {
    "SigridCode": {
      "command": "npx",
      "args": [
        "mcp-remote",
        "https://sigrid-says.com/mcp",
        "--header",
        "Authorization: Bearer <your_sigrid_token>",
        "--allow-http"
      ]
    }
  }
}

• Save → Verify server appears in tools list

Visual Studio

• Connect your GitHub account and open GitHub Copilot
• At the bottom left, below the chat box, select “Agent” mode
• Click on the + button to add a new MCP server
• When prompted, enter the name “SigridCode”, then add the URL as https://sigrid-says.com/mcp
• Choose “Additional headers” and add: Authorization: Bearer <your_sigrid_token>
• After saving/closing the window, if the token is valid, verify server appears in tools list

Windsurf

• Install Node
• Open MCP settings
• Add configuration:

{
  "mcpServers": {
    "SigridCode": {
      "command": "npx",
      "args": [
        "mcp-remote",
        "https://sigrid-says.com/mcp",
        "--header",
        "Authorization: Bearer <your_sigrid_token>",
        "--allow-http"
      ]
    }
  }
}

• Restart Windsurf

Claude Code (manual)

claude mcp add --transport http SigridCode https://sigrid-says.com/mcp --header "Authorization: Bearer <your_sigrid_token>"

Alternatively, using a proxy:

claude mcp add SigridCode -- npx mcp-remote https://sigrid-says.com/mcp --header "Authorization: Bearer <your_sigrid_token>" --allow-http

Replace <your_sigrid_token> with your actual Sigrid API token. Restart Claude Code after adding.

OpenCode

• Create or edit opencode.json in your project root:

{
  "mcp": {
    "SigridCode": {
      "type": "remote",
      "url": "https://sigrid-says.com/mcp",
      "headers": {
        "Authorization": "Bearer <your_sigrid_token>"
      }
    }
  }
}

• Restart OpenCode

IntelliJ / PyCharm / WebStorm

Navigate to Tools > AI Assistant > Model Context Protocol (MCP) and add:

"mcpServers": {
    "CodeGuardrails": {
      "command": "npx",
      "args": [
        "mcp-remote",
        "https://sigrid-says.com/mcp",
        "--header",
        "Authorization: Bearer <your_sigrid_token>",
        "--allow-http"
      ]
    }

IBM Bob

Navigate to Settings (the cogwheel icon inside the Bob chat window) > MCP. Configure a global or project-specific MCP server:

{
    "mcpServers":
    {
        "CodeGuardrails": {
            "type": "streamable-http",
            "url": "https://sigrid-says.com/mcp",
            "headers": {
              "Authorization": "Bearer <your_sigrid_token>"
            }
        }
    }
}

Note the extra outer brackets required for the configuration to validate successfully. Save the configuration and verify the connection on the settings page.

Available tools reference

Troubleshooting