PlugWiki
lanhu-mcp logo

Lanhu MCP Server: Setup for Claude, Cursor and Other Apps

CommunityLocal

Lets Claude, Cursor and other assistants read your Lanhu requirement documents and UI designs, and export design assets.

This is a community server made by dsphper for Lanhu (蓝湖), the Chinese design and prototype handoff tool. Your assistant can read Axure requirement pages, look at UI design images with exact sizes and styles, and export slices and icons. It is not made by Lanhu.

It runs on your computer and signs in with your Lanhu browser cookie, so you clone the repo and run its install script first. The assistant needs a model that can read images. A shared team message board lets several people's assistants leave notes for each other.

Good for

  • Front-end developers who build screens from Lanhu designs
  • QA engineers who write test cases from Lanhu Axure prototypes
  • Product managers who want an assistant to check requirements against a prototype
  • Teams that want one shared message board for their AI assistants

Works in

12 should work

Tap an app to see how to set it up and any known problems.

Install lanhu MCP

Claude Code

  1. 1

    Copy this prompt.

    Add the lanhu MCP server to Claude Code. Run this command:
    
    claude mcp add lanhu -e LANHU_USER_ROLE=Developer -- /bin/bash <ABS_PATH>/run-stdio.sh
    
    Then check that it loaded and tell me if I need to sign in or restart.
  2. 2

    Paste it into Claude Code. It shows the change and asks before running it.

Requires Git, Python 3.10-3.13, Run bash easy-install.sh in the cloned repo.

Clone https://github.com/dsphper/lanhu-mcp, run bash easy-install.sh (Windows: easy-install.bat), paste your Lanhu cookie when asked, and answer n to starting the HTTP service. Then replace <ABS_PATH> with the full path of the cloned folder. This config is adapted from the README by PlugWiki.

Built from the package data we hold. Not yet tested by us.

Tools

  • Lists the pages of a Lanhu requirement prototype

    Opens a Lanhu Axure document link and returns its page list. The assistant must call this before it analyzes a requirement document.

    Try: “Look at this Lanhu requirement link and list its pages.”

    Write · lanhu_get_pages

  • Reads requirement pages as text and screenshots for analysis

    Pulls the text, notes and rendered images of chosen prototype pages in stages, from the developer, tester or quick-scan point of view. Pages and images are cached on disk after the first read.

    Try: “Analyze the order page of this Lanhu document for test cases.”

    Write · lanhu_get_ai_analyze_page_result

  • Zooms into one spot of a requirement page

    Finds a phrase or area on a rendered prototype page and returns a clear crop around it, so small fields stay readable. When a phrase matches several places it returns previews of each instead of guessing.

    Try: “Show me the area around the red box that mentions the refund limit.”

    Files · lanhu_inspect_requirement_region

  • Lists the requirement documents in a Lanhu project

    Returns every product document or prototype in a project, so you can pick one before you ask for its pages.

    Try: “What requirement documents are in this Lanhu project?”

    Network · lanhu_list_product_documents

  • Turns a Lanhu invite link into a project link

    Converts a Lanhu invite or share link into the project address and ids the other tools need.

    Try: “Open this Lanhu invite link and tell me which project it points to.”

    Network · lanhu_resolve_invite_link

  • Lists the UI design images in a Lanhu project

    Returns the design screens in a project with their ids. Call it first, then pass an id to the design tools.

    Try: “List the UI designs in this Lanhu project.”

    Write · lanhu_get_designs

  • Snapshots one UI design with a numbered map of its layers

    Fixes one design version, returns an image of it, and lists its layers with stable numbers. Use it before you zoom into parts of a design.

    Try: “Give me an overview of the login screen design.”

    Write · lanhu_get_design_overview

  • Crops part of a design with its exact styles and asset ids

    Returns a clear crop of a chosen area or layers of a design, with each layer's source styles and linked asset ids. Use it to get exact sizes, spacing, colors and fonts.

    Try: “Crop the header of this design and give me its sizes and colors.”

    Write · lanhu_inspect_design_region

  • Shows a whole design with a draft HTML and CSS reference

    Older tool that previews a full design and builds a draft HTML and CSS reference from its data. The generated code is only a hint, and the project prefers the overview and region tools for exact evidence.

    Try: “Show this design with a first HTML and CSS draft.”

    Write · lanhu_get_ai_analyze_design_result

  • Lists the slices and icons in a design for download

    Finds the cut-out images, icons and other assets in a design and downloads them to the server's data folder with readable names.

    Try: “Get the icons and slices from this design.”

    Write · lanhu_get_design_slices

  • Exports original design assets as a ZIP with a manifest

    Downloads the original design assets you pick, checks their size and hash, and delivers a ZIP with a manifest. It never upscales images.

    Try: “Export the three icons I selected as a ZIP.”

    Write · lanhu_export_design_assets

  • Posts a note to the team's shared message board

    Writes a message on the project's board that other people's assistants can read. A message can be a normal note, a task to look something up, a question, or shared know-how, and it can @mention teammates.

    Try: “Tell Li Si that the refund rule on the order page conflicts with the design.”

    Write · lanhu_say

Read from the source code

Permissions

high risk · 5 things to watch

High risk

The assistant can read your Lanhu documents and designs with your own login, save files on your computer, and post, edit or delete notes on a shared message board. Keep the Lanhu cookie private, and use the stdio setup rather than the HTTP server unless you need to share it.

What it can do

  • Read your data
  • Create and change things
  • Delete things
  • Read-only mode

Watch out for

  • It can delete message board posts

    The delete tool removes posted messages from the server's team board. It does not delete anything in Lanhu itself. github.com

  • HTTP mode listens on every interface with no sign-in

    The server starts in HTTP mode unless you pass --transport stdio or set MCP_TRANSPORT=stdio, and its default address is 0.0.0.0 on port 8000. It has no sign-in of its own, so anyone who can reach the port can use your Lanhu cookie to read your documents and post or delete board messages. Use the stdio setup, or set SERVER_HOST to 127.0.0.1. github.com

  • It uses your full Lanhu login cookie

    LANHU_COOKIE is your whole Lanhu browser session, not a limited token, so the server can open everything your account can and it works until the session expires. It is stored as plain text in the .env file in the cloned folder. Keep that file out of repos and log out of Lanhu to invalidate the cookie if it leaks. github.com

  • It saves pages, images and exports to disk

    Design and requirement downloads are cached in a data folder next to the project. Clear it when you finish with a project. github.com

  • Every call logs your name and role to disk

    Most tools save the caller's name and role, with first and last access times, in a JSON file in the data folder, which lanhu_get_members then lists to anyone using the same server. github.com

Security advisories

No published security advisories as of Nov 1, 2026.

Technical details
Talks tolanhuapp.com, dds.lanhuapp.com and axure-file.lanhuapp.com. Feishu webhook only if you set FEISHU_WEBHOOK_URL
Sign-in and keysYour Lanhu cookie goes in a local .env file or the LANHU_COOKIE variable.
Scopes
  • Whatever your Lanhu login can open

Questions

Is the Lanhu MCP server official?

No. It is a community project by dsphper, and it is not made by Lanhu.

How does it sign in to Lanhu?

It uses the cookie from your Lanhu browser session, set as LANHU_COOKIE in a .env file. The install script asks for it, and the repo has a short guide to find it.

What do I need to install first?

Git, and Python 3.10 to 3.13. The install script then creates its own environment, adds a Chromium browser for screenshots, and writes the .env file.

Do I need a model that can read images?

Yes. The README says it needs a model with image understanding, such as Claude, GPT, Gemini, Kimi, Qwen or DeepSeek. Text-only models do not work.

Can I run it without Docker?

Yes. The stdio setup in the install box needs no Docker and starts only when your app connects. The repo also ships a Docker setup that keeps an HTTP server running.

Is it read-only?

No. It can save files on your computer and post, edit and delete messages on its team board. It does not change your Lanhu documents or designs.

Does it work in English?

The repo has an English README. The tool descriptions inside the server mix Chinese and English, and its design and requirement tools are built for Lanhu projects.

Information

updated Nov 2, 2026

Publisher
Community
Transport
stdio, Streamable HTTP
Auth
token in env
License
MIT
Last commit
Oct 27, 2026
Open issues
41

GitHubChangelogIssues

Sources

Docs, issue trackers and discussions this page is built from. Issues, quotes and findings above link to their own source.

Show all 3
  1. github.com · github.com
  2. github.com · github.com
  3. github.com · github.com

Facts from the official MCP registry, GitHub, npm, PyPI and the Docker catalog. Updated Nov 2, 2026.