Lspace

The Lspace MCP Server implements the Model Context Protocol (MCP) to let AI agents and tools access and manage a unified, searchable knowledge base across multiple repositories. It enables you to generate, organize, and retrieve knowledge from raw documents and synthesized entries through MCP clients, delivering persistent context for your AI workflows.

How to use

You run the Lspace MCP Server locally and connect MCP clients like Cursor or Claude Desktop to it. The server exposes a programmatic interface that lets AI agents query, classify, organize, and summarize content stored in raw documents and knowledge bases. By integrating with your MCP clients, you can generate Wikipedia-like knowledge bases from your repositories, navigate document timelines, and perform AI-assisted knowledge discovery across multiple sources.

Key usage patterns you can adopt include:

• Connect an MCP client to the Lspace server and start a knowledge base generation workflow from raw inputs.
• Use AI orchestration to classify, organize, and summarize documents into topic pages with cross-references.
• Manage multiple repositories (local and GitHub) and maintain a synchronized knowledge base across them.
• Leverage the timeline to track operations on documents and their synthesis into the knowledge base.

How to install

Prerequisites you need before installation:
- Node.js (LTS version) including npm
- Git
- A GitHub Personal Access Token (PAT) if you plan to connect GitHub repositories via MCP

Step 1: Clone the Lspace server repository
```bash
git clone https://github.com/Lspace-io/lspace-server.git

Step 2: Enter the project directory

cd lspace-server

Step 3: Install dependencies

npm install

Step 4: Build the project (compiles TypeScript to JavaScript in the dist folder)

npm run build

The MCP server entry point after build is lspace-mcp-server.js in the project root.

Step 5: Configure environment and repositories (details below) and then start using the MCP server with your MCP client.

## Configuration and usage notes

Environment configuration
- Create a local environment file for the server by copying the example:
```bash
cp .env.example .env```- Edit the new` .env `file and set your` OPENAI\_API\_KEY` and other variables as needed.

Repository configuration
- Create a local configuration file to tell Lspace which repositories to manage and supply credentials (e.g., GitHub PATs). Copy the example configuration and edit it:
```bash
cp config.example.json config.local.json```- In` config.local.json`, add your GitHub PATs under` credentials.github\_pats `and define the repositories you want Lspace to manage under` repositories`.

Connecting MCP clients
- The MCP entry point is the `lspace-mcp-server.js` script built in the previous steps. You’ll configure MCP clients (e.g., Cursor or Claude Desktop) to invoke this script using the Node runtime, pointing to the correct path for your local setup.

## Security and credentials

GitHub PATs are required for Lspace to access your GitHub repositories. Create a PAT with the `repo` scope to enable cloning, reading, writing, and pushing changes. Store PATs securely and reference them via `credentials.github_pats` in your local configuration.

OpenAI keys and other sensitive values should be kept in your environment file and not committed to version control. Use environment-specific overrides only when necessary for your setup.

## Examples and tips

- Use a dedicated mcp.json or claude-like config to point clients to the MCP server, with `command` set to `node` and `args` containing the path to `lspace-mcp-server.js`.

## Available tools

### kb\_generation

Generates a knowledge base from raw documents and synthesizes topic pages that summarize content across repositories.

### ai\_orchestration

Coordinates AI agents to perform classification, organization, and summarization tasks on documents and knowledge bases.

### document\_classification

Classifies and organizes documents into taxonomies and topics to support coherent knowledge base structure.

### timeline\_tracking

Records and exposes the timeline of document operations and transformations within the knowledge base.

### knowledge\_base\_synthesis

Creates synthesized, Wikipedia-like knowledge pages that reference source documents and cross-link related entries.

### search\_and\_link

Enables cross-referencing and searching across raw inputs and knowledge base entries.