evaluate_readmeLABSCO SUMMARY
It judges presentation, not content.
This will not tell you whether your README explains the project. It will tell you, consistently and across a whole directory of them, whether the header renders well, whether the badges are where readers look, and whether the two language versions match — the things that are obvious in one repository and invisible across thirty. Read it as a house-style checker for a set of projects rather than as documentation review, and note the checklist is opinionated: centred headers, SVG banners, English and Japanese badges.
What it is
An MCP server that evaluates README files in a project directory against a presentation checklist: whether there is a header image and how good it is, whether the title and badges are centred, and whether English and Japanese documentation badges are present and correctly formatted. It returns a score with the individual findings behind it.
What you get
- Every README in a project directory evaluated from one call, with the path to each —
evaluate_readme - A structured result rather than prose: whether a header image exists, its quality broken into gradient, animation, rounded corners, English text and project-specificity, whether the header image, title and badges are centred, whether English and Japanese badges exist and use the right format, a numeric score, and a list of suggestions
- SVG header analysis that reads the file, so "is this image actually about this project" is answered from its contents rather than from its filename
Requirements
Nothing external — no account, no key, no network; it reads files from a path you give it. Node.js 18+ and npm or yarn. There is no published package: clone, npm install, npm run build, and point your client at build/index.js. Header images need to be SVG files in the project's assets/ directory with valid XML for the analysis to find them, and badges are recognised in shields.io format.
Setup effort
Build from source
USE CASES
Bringing a set of repositories to the same standard
Run the evaluation across the project and compare scores, so the ones that fell behind are named rather than guessed at.
Finding out why a header looks wrong
Read the header-image quality breakdown, which says which of the specific properties is missing rather than that the image is poor.
Checking bilingual documentation stays in step
Let it verify that both language badges exist, are centred and use the expected format, before you publish.
Getting a concrete edit list
Take the suggestions array as the to-do rather than reading the score and deciding what it meant.
1 tool
Listed in project docs. Your client may show a different list.
BEFORE YOU INSTALL
What to know first
Tools you get
1 tool named in the project's documentation.
Requires
Node.js 18+
Where it runs
The documented client runs the server on your machine.
INSTALLATION
Connect your client
You clone and build this one yourself.
Server setup
git clone https://github.com/Sunwood-ai-labs/documind-mcp-server
cd documind-mcp-server
npm install
npm run buildMCP Clients
Any MCP client
{
"mcpServers": {
"documind-mcp-server": {
"command": "<YOUR_PATH_TO_DOCUMIND_MCP_SERVER>/build/index.js"
}
}
}