Setting Up MCP in Continue: Complete Configuration Guide

What is Continue?

Continue is an open source AI code assistant that integrates with VS Code and JetBrains IDEs. Unlike proprietary alternatives, Continue lets you choose your own models (OpenAI, Anthropic, local models via Ollama, and more) while keeping your data private.

With MCP integration, Continue can access external APIs, databases, file systems, and other tools—making it a good fit for developers who want flexibility and control.

MCP in Continue Overview

Continue's MCP implementation uses YAML configuration files, offering a more readable alternative to JSON. Key characteristics:

• ●Config location: .continue/mcpServers/*.yaml (standalone files)
• ●Alternative: mcpServers array in config.yaml
• ●Format: YAML (or JSON import from Claude/Cursor)
• ●Mode: MCP tools only available in Agent mode

YAML Configuration

Continue supports two ways to configure MCP servers:

Option 1: Standalone YAML Files

Create individual YAML files in the .continue/mcpServers/ directory:

# .continue/mcpServers/filesystem.yaml

name: Filesystem Server
version: 1.0.0
schema: v1

transport:
  type: stdio
  command: npx
  args:
    - "-y"
    - "@modelcontextprotocol/server-filesystem"
    - "/Users/username/projects"

Option 2: In config.yaml

Add servers to your main config.yaml file:

# .continue/config.yaml

mcpServers:
  - name: filesystem
    transport:
      type: stdio
      command: npx
      args:
        - "-y"
        - "@modelcontextprotocol/server-filesystem"
        - "/Users/username/projects"
  
  - name: github
    transport:
      type: stdio
      command: npx
      args:
        - "-y"
        - "@modelcontextprotocol/server-github"
      env:
        GITHUB_PERSONAL_ACCESS_TOKEN: "${{ secrets.GITHUB_TOKEN }}"

Configuration Fields

* Required in standalone YAML files, optional when defined in config.yaml

Transport Types

Continue supports three transport methods for MCP servers:

Importing JSON Configs

If you already have MCP servers configured in Claude Desktop or Cursor, you can import their JSON configurations directly into Continue.

Import from Claude Desktop

Add the mcpServersJSONfield to your config.yaml pointing to Claude's config:

# .continue/config.yaml# Import from Claude Desktop (macOS)# Or import from Cursor# Or import from Cursor# Or import from Cursor# Or import from Cursorort from Cursor
mcpServersJSON: ~/.cursor/mcp.json

Combined Configuration

You can combine imported JSON with YAML-defined servers:

# .continue/config.yaml# Import existing servers from Claude# Add Continue-specific servers# Add Continue-specific servers# Add Continue-specific serverstinue-specific servers
mcpServers:
  - name: custom-server
    transport:
      type: stdio
      command: python
      args:
        - /path/to/my-custom-server.py

Managing Secrets

Continue provides secure secrets management using the ${{ secrets.NAME }} syntax:

Setting Secrets

Secrets are stored securely in Continue's configuration:

1. Open Continue settings (Cmd+Shift+P → “Continue: Open Settings”)
2. Navigate to the Secrets section
3. Add your secret with a name (e.g., GITHUB_TOKEN)
4. Reference it in config as ${{ secrets.GITHUB_TOKEN }}

Example with Secrets

# .continue/mcpServers/github.yaml

name: GitHub Integration
version: 1.0.0
schema: v1

transport:
  type: stdio
  command: npx
  args:
    - "-y"
    - "@modelcontextprotocol/server-github"
  env:
    GITHUB_PERSONAL_ACCESS_TOKEN: "${{ secrets.GITHUB_TOKEN }}"

Configuration Examples

Filesystem Server

# .continue/mcpServers/filesystem.yaml

name: Filesystem
version: 1.0.0
schema: v1

transport:
  type: stdio
  command: npx
  args:
    - "-y"
    - "@modelcontextprotocol/server-filesystem"
    - "/Users/username/projects"
    - "/Users/username/documents"

PostgreSQL Database

# .continue/mcpServers/postgres.yaml

name: PostgreSQL
version: 1.0.0
schema: v1

transport:
  type: stdio
  command: npx
  args:
    - "-y"
    - "@modelcontextprotocol/server-postgres"
  env:
    POSTGRES_CONNECTION_STRING: "${{ secrets.DATABASE_URL }}"

Remote SSE Server

# .continue/mcpServers/remote-api.yaml

name: Remote API
version: 1.0.0
schema: v1

transport:
  type: sse
  url: https://api.example.com/mcp/sse

Python Custom Server

# .continue/mcpServers/custom.yaml

name: My Custom Server
version: 1.0.0
schema: v1

transport:
  type: stdio
  command: python
  args:
    - /path/to/my-mcp-server.py
  env:
    API_KEY: "${{ secrets.MY_API_KEY }}"
    DEBUG: "true"

Using Agent Mode

MCP tools are only available in Continue's Agent mode. Here's how to use them:

Switching to Agent Mode

1. Open the Continue panel in VS Code or JetBrains
2. Click the mode selector (usually shows “Chat”)
3. Select “Agent” from the dropdown
4. MCP tools will now be available for the agent to use

How Agent Uses MCP Tools

In Agent mode, Continue automatically decides when to use MCP tools based on your prompt:

• ●The agent analyses your request and available tools
• ●It plans a sequence of actions (which may include MCP tool calls)
• ●You can approve or reject each tool call
• ●The agent processes results and continues until the task is complete

Troubleshooting

References & Further Reading