opencode-mcp

OpenCode MCP Servers

📚 Official Docs: For the latest information, always refer to the official documentation: https://opencode.ai/docs/mcp-servers/

Add local and remote MCP tools to OpenCode using the Model Context Protocol (MCP). MCP is a standard protocol that lets you connect external tool servers to AI applications. OpenCode supports both local servers (child processes) and remote servers (HTTP). Once added, MCP tools are automatically available to the LLM alongside OpenCode's built-in tools.

---

Context Caveat

When you use an MCP server, it adds tokens to the context. Every tool the MCP server exposes is described in the context window, which consumes tokens. This can quickly add up if you have a lot of tools from multiple MCP servers. Be careful with which MCP servers you enable.

• Certain MCP servers, like the GitHub MCP server, tend to add a lot of tokens and can easily exceed the context limit.
• Each additional MCP server multiplies the tool definitions injected into context.
• Monitor your context window usage and be selective about which servers you enable.

---

Enable / Disable

MCP servers are defined in your OpenCode Config (opencode.json or opencode.jsonc) under the mcp key. Each server gets a unique name used to reference it in prompts and tool management.

{
  "$schema": "https://opencode.ai/config.json",
  "mcp": {
    "name-of-mcp-server": {
      // ... server config ...
      "enabled": true
    },
    "name-of-other-mcp-server": {
      // ... server config ...
    }
  }
}

• Set "enabled": true to explicitly enable a server.
• Set "enabled": false to temporarily disable a server without removing its config.
• If enabled is omitted, the server is enabled by default.

---

Local MCP Servers

Local servers run as child processes on your machine. Use "type": "local" with a command array.

Full Configuration

{
  "$schema": "https://opencode.ai/config.json",
  "mcp": {
    "my-local-mcp": {
      "type": "local",
      "command": ["npx", "-y", "my-mcp-command"],
      "enabled": true,
      "environment": {
        "MY_ENV_VAR": "my_env_var_value"
      },
      "timeout": 5000
    }
  }
}