Lanhu MCP Server
MCP server for the Lanhu design service: the agent reads Axure prototypes and designs, exports slices and shares a team board
- 2,572 ★ on the repository
- Last commit
- MIT
Broad access
Why this level
- The Lanhu account cookie gives the agent every project that account can reach
- The team board stores messages and the sample config binds the server to 0.0.0.0, so it should stay on an internal network
- Board tasks ask other people's agents to run code and database queries, which the docs limit to read-only
Install
- Works in
Manual install
git clone https://github.com/dsphper/lanhu-mcp.git && cd lanhu-mcp && cp .env.example .env && docker compose up -d --buildPut your `LANHU_COOKIE` into `.env` before starting. The server opens at `http://localhost:8000/mcp`.
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- The Lanhu account cookie gives the agent every project that account can reach
- The team board stores messages and the sample config binds the server to 0.0.0.0, so it should stay on an internal network
- Board tasks ask other people's agents to run code and database queries, which the docs limit to read-only
This is third-party code. Review the repository files before installing.
What it does
Lanhu MCP Server connects an agent to Lanhu (lanhuapp.com), a Chinese design handoff service. The server works as your account through a cookie: it hands the agent prototypes and designs and runs a team board. The agent gets Axure prototype pages as a list and analyzes them from a developer, tester or quick-overview angle. For long pages the server returns a full screenshot and overlapping tiles. The `lanhu_inspect_requirement_region` tool resolves vague hints such as "the red box" by text, block ID or coordinates. For interface designs the server pins a design version and returns an overview, crops by layer ID, styles, sizes, colors and a reference HTML and CSS. The `lanhu_export_design_assets` tool downloads original slices with size and hash checks. On the team board the `lanhu_say*` tools store notes, knowledge and tasks that agents of different people read from the shared server. Mentions can be sent to Feishu. The server runs in Docker, from source, or on demand over stdio.
Who it is for. For teams that pass requirements and designs through Lanhu and want Cursor, Claude Code or another agent to read them directly.
Good fit when
- Requirements sit in an Axure prototype on Lanhu and a developer needs a field and logic breakdown
- A tester needs scenarios, boundary values and field checks from the prototype
- A front-end developer needs sizes, colors, fonts and original slices from a Lanhu design
- The team needs to leave notes and knowledge for agents on a shared server and mention colleagues in Feishu
Not a fit when
- Your designs live in Figma: Framelink MCP for Figma fits better
- Your agent runs a text-only model: the server needs a model that can read images
- The server would have to face the public internet: keep it on an internal network or behind a firewall
- You cannot hand the Lanhu account cookie to third-party software
Example request
Use the MCP to look at this Lanhu prototype and write a breakdown for a developerMCP
- Transport
- http, stdio
- Authentication
- API key
Environment variables
LANHU_COOKIErequired, secret- Cookie from your Lanhu web session
FEISHU_WEBHOOK_URLsecret- Feishu webhook URL for mention notifications
LANHU_USER_NAME- User name for board entries in stdio mode
LANHU_USER_ROLE- User role: Developer, Tester, Product and others
Limitations
Python 3.10 or newer is required, plus Playwright Chromium for some tools. The `LANHU_COOKIE` variable from your Lanhu web session is mandatory. Documentation and prompts are mostly Chinese, with an English README. Reference HTML and CSS from a design must be checked against the original markup. Feishu notifications work only with `FEISHU_WEBHOOK_URL` set. It depends on lanhuapp.com being reachable from your network.
How to disable. Remove the `lanhu` entry from the client's MCP configuration. `docker compose down` stops the container, and board data sits in the `DATA_DIR` directory.
FAQ
How do I connect Lanhu to Claude Code or Cursor?
Start the server with `docker compose up -d --build` and add `http://localhost:8000/mcp?role=Developer&name=YourName` to the client's MCP configuration.
Where do I get `LANHU_COOKIE`?
Sign in to Lanhu in a browser, open developer tools and copy the Cookie header from a request. A step-by-step guide is in `GET-COOKIE-TUTORIAL.md`.
Which models work?
Vision-capable models: Claude, GPT, Gemini, Kimi, Qwen or DeepSeek.
What does the team board do?
Agents leave knowledge and task notes and read notes from others. `lanhu_say_list` finds messages by topic, and mentions go to Feishu when `FEISHU_WEBHOOK_URL` is set.
Can I run the server without Docker?
Yes, from source or on demand over stdio. The `run-stdio.sh` and `run-stdio.bat` scripts read `.env` and start the server, as described in `LOCAL-STDIO.md`.
Discussion
No questions or reviews yet.
Related
MCP serversFrom a Lanhu design link an agent gets HTML with CSS, design tokens, layers and slices, with no manual export from the editor.
A design studio for people and agents: dsa CLI, MCP endpoint and skill to edit layouts, slides, reports and 3D together.
An MCP server that gives the agent simplified Figma layout data from a link so it can build the screen more accurately than from a screenshot
Official Webflow MCP server: CMS, pages and on-canvas elements in the Designer via the Data API
An MCP server that lets an agent read and edit Figma designs: colors, typography, components and export
Figma MCP Server (Antonytm)
Figma MCP Server
A community MCP server with a Figma plugin: the agent creates frames, text and components in the open file, not just reads them
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Authentication | Repository stars | Main difference |
|---|---|---|---|---|
| Lanhu MCP Server (this card) | no | API key | 2,572 | |
| Framelink MCP for Figma | no | API key | 16,202 | works with Figma layouts and does not read Lanhu prototypes |
| Figma MCP Server | yes | OAuth | 2,179 | the official server for Figma designs |
README in short
The docs cover install through Docker and from source, client setup and a list of 17 tools for requirements, designs and the board. Separate documents explain running without Docker, slice export and the design context. The release notes reach version 1.8.7.