DeepSeek MCP Server for VS Code: Setup & Config

What you can do with DeepSeek in VS Code

Pick it for explicit DeepSeek V4 controls plus model, balance, and in-memory conversation utilities.

• Generate single-turn or multi-turn DeepSeek chat responses
• Inspect available models and account balance

Before you start: DeepSeek in VS Code

• npx
• A DeepSeek API key for local mode
• An MCP-compatible client

This is the local stdio launch. A hosted HTTP alternative is documented at https://deepseek-mcp.ragweld.com/mcp and requires its own bearer token; it is not needed for this local configuration.

DeepSeek VS Code configuration

1. Choose VS Code configuration scope

   For DeepSeek MCP Server, open Workspace: .vscode/mcp.json. Merge into servers; preserve existing inputs.

2. Add DeepSeek MCP Server to VS Code

   Use the deepseek-mcp-server entry below in servers. Keep DeepSeek MCP Server's documented stdio connection fields and your other VS Code entries.

3. Check DeepSeek MCP Server in VS Code

   DeepSeek MCP Server: Start in VS Code’s MCP controls; inspect output and trust prompts.

{
  "servers": {
    "deepseek-mcp-server": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "deepseek-mcp-server"
      ],
      "env": {
        "DEEPSEEK_API_KEY": "<DeepSeek provider API key>"
      }
    }
  }
}

DeepSeek tools available to VS Code

DeepSeek's stored tool list documents 7 tools; compare it with VS Code's connected tool list, because the stored list is not a live check.

chat_completion

Look for chat_completion among VS Code's agent-mode tools after starting DeepSeek through the MCP controls.

completion

Look for completion among VS Code's agent-mode tools after starting DeepSeek through the MCP controls.

Generate V4 Pro fill-in-the-middle text from a prompt and suffix.

create_response

Look for create_response among VS Code's agent-mode tools after starting DeepSeek through the MCP controls.

Create a stateless Responses API result with optional instructions, reasoning controls, and tools.

list_models

Look for list_models among VS Code's agent-mode tools after starting DeepSeek through the MCP controls.

get_user_balance

Look for get_user_balance among VS Code's agent-mode tools after starting DeepSeek through the MCP controls.

Read account balance and availability without changing account state.

reset_conversation

Look for reset_conversation among VS Code's agent-mode tools after starting DeepSeek through the MCP controls.

Clear one conversation history held by the current MCP process.

list_conversations

Look for list_conversations among VS Code's agent-mode tools after starting DeepSeek through the MCP controls.

DeepSeek: full stored tool descriptions

Troubleshooting DeepSeek in VS Code

• Make npx deepseek-mcp-server (Node.js/npm) available in the workspace, container, or remote environment hosting DeepSeek MCP Server. Start through VS Code's MCP controls; inspect output and trust prompts.
• For DeepSeek MCP Server, keep DEEPSEEK_API_KEY out of version control; VS Code documents input variables and secret handling.
• DeepSeek MCP Server uses stdio here: VS Code must keep the configured process running; protocol output belongs on stdout and diagnostics on stderr.
• Known limitation: Conversation history is limited to the current process memory
• Known limitation: Provider credentials or a hosted bearer token are required
• Known limitation: Responses API tool appears in structured data but not the README tool list
• General VS Code troubleshooting lives in the /clients/vscode guide.

Frequently asked questions