JetBrains IDE MCP Server
Connects to DataGrip's MCP Server, exposing database-focused tools for listing connections, executing SQL queries, previewing table data, introspecting schemas, and listing schema objects.
Connects to IntelliJ IDEA's MCP Server so the agent can use IDEA's project build/run, symbol and text search, code analysis, refactoring, inspections, debugger, database, VCS, and terminal tools.
Integrates with JetBrains IDEs that have the MCP Server plugin enabled, exposing the IDE's build/run, code search, refactoring, inspection, debugger, database, VCS, and terminal capabilities to the agent.
Connects to PyCharm's MCP Server, exposing PyCharm's build/run, code search, refactoring, inspection, debugger, database, VCS, and terminal tools.
Connects to WebStorm's MCP Server, letting the agent use WebStorm's build/run, code search, refactoring, inspection, debugger, database, VCS, and terminal tools.
This server does not currently have a deployable release on Glama. Follow the repository's setup instructions to install and configure it for your MCP client, including any required credentials and local dependencies.
dsh-jetbrains-mcp
JetBrains IDE MCP Server for DeepSeek Harness — the IDE's own build, refactoring, debugger, database and Git tools, in the agent's tool list.
English | 中文
What it gives the agent
Once installed and enabled, the model sees IntelliJ IDEA's MCP tools under a stable namespace:
Group | Tools |
Build & run |
|
Code understanding |
|
Quality |
|
Editing |
|
Debugger | the |
Database |
|
VCS |
|
Terminal |
|
Names are mcp__jetbrains__<tool>, the same shape Claude Code and Codex use, so a tool reference in chat history or a permission rule keeps working across restarts.
Plus one tool of its own: jetbrains_mcp_doctor, which explains what is wrong when the mcp__jetbrains__* tools are missing.
Related MCP server: dsh-trio
Requirements
IntelliJ IDEA, PyCharm, WebStorm, DataGrip or any other JetBrains IDE, 2025.2 or newer (any JetBrains IDE that bundles the MCP Server plugin) with Settings | Tools | MCP Server → Enable MCP Server switched on. The IDE may use any port; this bundle discovers it.
DeepSeek Harness 0.2.0 or newer with the in-box
@deepseek-ai/dsh-mcp-clientbridge (DSH Desktop, ordsh web).No Node.js on
PATHis required: the adapter runs on the host's own runtime.
Install
From the plugin market (recommended). Open Plugins → Discover in DSH, search for dsh-jetbrains-mcp, and install it.
From a terminal.
dsh plugin --profile <profile> add dsh-jetbrains-mcpReplace <profile> with the profile your host uses (desktop for DSH Desktop's own profile is managed by the app — install through Settings → Plugins there). Restart the host afterwards.
From a checkout of this package.
dsh plugin --profile <profile> add /absolute/path/to/dsh-jetbrains-mcpUsage
Most IDE tools take a projectPath argument — the directory opened in the IDE:
use mcp__jetbrains__build_project with projectPath "F:/work/my-service"Without it the IDE answers "Unable to determine the target project": the MCP server is IDE-wide, so the project is not implied by the connection. Everything else behaves like any other tool.
How it works
DSH agent
└─ tool call mcp__jetbrains__build_project
└─ @deepseek-ai/dsh-mcp-client (in-box, mounted by this bundle's patch row)
└─ bin/jetbrains-mcp-bridge.mjs (this package, zero dependencies)
│ stdio ⇄ HTTP+SSE
└─ IntelliJ IDEA MCP Server http://127.0.0.1:<port>/sseJetBrains IDEs' MCP Server exposes the legacy HTTP+SSE transport (GET /sse, then POST /message?sessionId=…), while DSH's in-box MCP client speaks stdio or streamable-http. The adapter in bin/ is the missing half: a stdio MCP server for DSH and an SSE MCP client for the IDE. It is a few hundred lines of Node built-ins — no SDK copy, no second protocol implementation beyond this translation.
The IDE's port is read from %APPDATA%\JetBrains\*\options\mcpServer.xml (fallback 64342) on every launch, so changing the port inside the IDE needs no edit here.
Row paths are computed by the Loader at mount time (!!js expressions over the entry's own context), so nothing is written into your profile configuration and no absolute path is baked into the package.
Overrides
For debugging only; set them in the environment the host runs in:
Variable | Effect |
| Full SSE URL, e.g. |
| Node executable to run the adapter |
| Absolute path of the adapter script |
Troubleshooting
Start with jetbrains_mcp_doctor — it reports the adapter path it found, the command the bridge row runs, every JetBrains product that has MCP settings and on which port, a live probe of that endpoint, and the registered mcp__jetbrains__* tools.
Symptom | Cause |
No | The row |
| No IDE settings — the MCP Server plugin is not installed/enabled, or no JetBrains IDE has run on this account |
Probe reports a timeout or an error | The IDE is closed, or its MCP server is disabled in Settings → Tools → MCP Server |
Tools exist but every call answers "Unable to determine the target project" | Pass |
Tools disappear when the IDE closes | Expected: the session dies with the IDE and the row reconnects on its own when it is back |
Security
The adapter connects to
127.0.0.1only.The IDE's Brave Mode (
enableBraveMode, on by default in some versions) makes the IDE execute MCP tool calls without asking again in the IDE. If you do not want that, turn Brave Mode off in the IDE's MCP settings; the agent's own approval flow still applies on the DSH side.Installing this plugin means an agent with tool access can build, run, refactor, debug and run SQL through your IDE. Treat it like handing over the IDE.
License
MIT — see LICENSE.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
This server cannot be deployed
Maintenance
Related MCP Connectors
Human-input bridge for AI agents with voice-first answer links, MCP tools, and HTTP APIs.
Talk to your public-facing AI from any MCP client — Claude, ChatGPT, Cursor, Cline, Windsurf.
Connect ChatGPT, Claude, Cursor or VS Code to your Windows PC: files, terminal, browser, apps.
Email inboxes, webhook URLs, mock APIs and 170+ tools for coding agents over one MCP connection.
Related MCP Servers
- AlicenseAqualityBmaintenanceBridges MCP clients to the DeepSeek API for chat completions and model discovery, with an optional locked-down bridge that delegates prompts to local CLI harnesses like Claude, Codex, or opencode.446 npmMIT
- AlicenseNot gradedqualityCmaintenanceTurns DeepSeek Harness into an MCP server with tools for session management, agent execution, resources, and OAuth, plus browser automation and GitHub/GitLab integration.30 npm20MIT
- AlicenseNot gradedqualityCmaintenanceBridges DeepSeek Harness ACP services to MCP stdio hosts like Codex, enabling session management, model selection, background tasks, cancellation, and subagent observation.MIT
- AlicenseNot gradedqualityCmaintenanceLets any MCP-capable AI coding harness route code reviews and codebase questions through DeepSeek's web chat, exposing review, ask, and status tools with read-only repository access by default.1MIT