1. Home
  2. Servers
  3. AI Models & Agents
  4. DeepSeek MCP Server

Server profile

DeepSeek MCP Server

DeepSeek MCP Server exposes DeepSeek V4 generation and account utilities to MCP clients. It offers chat, fill-in-the-middle completion, model discovery, balance checks, and in-process conversation management.

View repositoryInstall & configure

Server facts

DeepSeek MCP Server

Tech

Runtime / language
node
Transports
stdio
Remote available
No
Authentication
Not specified

Activity

GitHub stars
353
Last commit
Data as of

Tags

  • Chat Completions
  • Conversation Memory
  • Deepseek
  • Fill In The Middle

Overview

DeepSeek MCP Server connects MCP clients to DeepSeek V4 generation and account operations. The documented chat models are deepseek-v4-pro and deepseek-v4-flash, while fill-in-the-middle generation uses V4 Pro. Chat defaults to the Flash model and supports thinking controls, reasoning effort, structured JSON output, function tools, log probabilities, streaming, and conversation memory. The structured input also lists a stateless Responses API tool, although the README's own enumerated tool list does not include that entry. This distinction matters when comparing the supplied catalog with the documented server surface. The README explicitly excludes speculative image, video, upload, monitoring, and automatic model-substitution features.

Supporting tools retrieve the live model catalog and the account balance, helping clients select models and inspect account availability. Conversation controls list or clear histories stored in the current MCP process; they should not be mistaken for a separately described persistent database. Local setup uses npx with a DeepSeek API key, and a versioned Docker image is also documented. A hosted MCP endpoint is provided with bearer-token authentication and examples for several clients. Optional environment settings control the provider base URL, timeout, default model, HTTP server behavior, and message-history limit. Local stdio and hosted HTTP access are separate setup choices, with distinct credentials shown for provider access and remote authentication. The project uses the MIT license.

Install & configure

Requirements

  • npx
  • A DeepSeek API key for local mode
  • An MCP-compatible client
Documented launch / install command
npx -y deepseek-mcp-server

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.

Config file · macOS: ~/Library/Application Support/Claude/claude_desktop_config.json

Windows: %APPDATA%\Claude\claude_desktop_config.json

Use Settings → Developer → Edit Config when available. Do not assume a Linux configuration path; check the client documentation for your platform. Remote servers use the app’s connector flow when supported, not the local stdio file shown here. Merge this example with existing settings; do not overwrite unrelated servers.

Claude Desktop configuration
{
  "mcpServers": {
    "deepseek-mcp-server": {
      "command": "npx",
      "args": [
        "-y",
        "deepseek-mcp-server"
      ],
      "env": {
        "DEEPSEEK_API_KEY": "<DeepSeek provider API key>"
      }
    }
  }
}

Read the Claude Desktop setup guide

Config file · Project: .cursor/mcp.json

User: ~/.cursor/mcp.json

The project path is relative to the workspace root. The tilde denotes your user home directory. Use Cursor’s MCP settings to confirm which configuration and scope your installed version is reading. Merge this example with existing settings; do not overwrite unrelated servers.

Cursor configuration
{
  "mcpServers": {
    "deepseek-mcp-server": {
      "command": "npx",
      "args": [
        "-y",
        "deepseek-mcp-server"
      ],
      "env": {
        "DEEPSEEK_API_KEY": "<DeepSeek provider API key>"
      }
    }
  }
}

Read the Cursor setup guide

Config file · Workspace: .vscode/mcp.json

For user scope, use the “MCP: Open User Configuration” command if available. Do not guess a user configuration path across operating systems, profiles, containers, or remote workspaces; check the client documentation. Merge this example with existing settings; do not overwrite unrelated servers.

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

Read the VS Code setup guide

Config file · macOS/Linux: ~/.codeium/windsurf/mcp_config.json

Windows: %USERPROFILE%\.codeium\windsurf\mcp_config.json

Open the MCP configuration through Windsurf’s settings when possible to confirm the active file. Paths can differ in managed or future releases; check the client documentation if this location is not present. Merge this example with existing settings; do not overwrite unrelated servers.

Windsurf configuration
{
  "mcpServers": {
    "deepseek-mcp-server": {
      "command": "npx",
      "args": [
        "-y",
        "deepseek-mcp-server"
      ],
      "env": {
        "DEEPSEEK_API_KEY": "<DeepSeek provider API key>"
      }
    }
  }
}

Read the Windsurf setup guide

Config file · Project scope: .mcp.json at the repository root

The project file uses mcpServers. For user scope, use claude mcp add --scope user rather than assuming .mcp.json in your home directory is the user configuration. Check the client documentation for managed settings and storage details. Merge this example with existing settings; do not overwrite unrelated servers.

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

Read the Claude Code setup guide

Examples are translated from the published configuration. Supply your own credentials and verify requirements in the upstream README. Remote connectors are shown only where a supported configuration format is known.

Tools

Source: repository tool metadata. Tool descriptions are a documented snapshot, not a live capability check.

ToolWhat it does
chat_completionGenerate DeepSeek V4 chat responses from a message or conversation history.
completionGenerate V4 Pro fill-in-the-middle text from a prompt and suffix.
create_responseCreate a stateless Responses API result with optional instructions, reasoning controls, and tools.
list_modelsRetrieve available DeepSeek model identifiers.
get_user_balanceRead account balance and availability without changing account state.
reset_conversationClear one conversation history held by the current MCP process.
list_conversationsList conversation identifiers currently held in process memory.

Names, permissions, and defaults can change. Check the current upstream documentation before granting access.

Use cases

  • Generate single-turn or multi-turn DeepSeek chat responses
  • Perform fill-in-the-middle text completion with V4 Pro
  • Inspect available models and account balance
  • List or reset conversation histories in the active process

Strengths

  • Includes chat, completion, and provider account utilities
  • Documents both local stdio and hosted HTTP setups
  • Exposes explicit model and thinking controls

Limitations

  • Conversation history is limited to the current process memory
  • Provider credentials or a hosted bearer token are required
  • Responses API tool appears in structured data but not the README tool list

Frequently asked questions

The README names deepseek-v4-flash as the chat default; fill-in-the-middle completion defaults to deepseek-v4-pro.

No. The README explicitly says automatic model substitution is not included.

Only in-memory histories in the current MCP process are described.

Local mode uses DEEPSEEK_API_KEY. The hosted endpoint uses a bearer token shown as DEEPSEEK_MCP_AUTH_TOKEN in client examples.

Independent listing, not an affiliation or security review. Grant the least access needed and verify the upstream instructions before running code.

Related MCP servers

MiniMax MCP

AI Models & Agents

MiniMax MCP exposes MiniMax media APIs to compatible AI clients for speech, voice work, images, video, and music. It supports local stdio use and SSE deployment, with region-specific API credentials and configurable resource output.

pythonstdio1,590 GitHub stars

ElevenLabs MCP

AI Models & Agents

ElevenLabs MCP brings the vendor's speech and audio services into assistant workflows. This official integration supports voice work and transcription, with outputs delivered as files, MCP resources, or both.

pythonstdio1,535 GitHub stars

Just Prompt

AI Models & Agents

Just Prompt lets an MCP assistant consult several language models and preserve their answers. Alongside parallel prompting, it offers a panel workflow in which a designated model makes a decision after reviewing other responses.

pythonstdio739 GitHub stars

mcp-sequentialthinking-tools

AI Models & Agents

mcp-sequentialthinking-tools records stepwise reasoning with session history, branches, and revision metadata. It validates model-authored tool plans against a supplied tool list rather than discovering or choosing tools itself.

nodestdio587 GitHub stars