MCP Lucene Server

Description

The MCP Lucene Server is a Java-based implementation of the Model Context Protocol (MCP) designed to provide efficient search and retrieval capabilities using Apache Lucene. This server allows you to manage and query documents, leveraging Lucene's powerful indexing and search features. It is built using Spring Boot for easy setup and deployment.

Features

• MCP Compliance: Implements the core Model Context Protocol.

• Lucene-Powered: Utilizes Apache Lucene for full-text search and indexing.

• RESTful API: Provides a RESTful API for interacting with the server.

• Document Management:

  • Upsert: Add or update documents in the Lucene index.

  • Delete: Delete documents from the Lucene index.

  • List: Retrieve a list of documents from the index.

• Querying:

  • Supports complex queries using the Lucene query syntax.

  • Filtering: Filter queries based on document metadata.

• Status: Check the server status.

• Spring Boot: Built with Spring Boot for easy setup and deployment.

• Dockerization: Includes instructions for containerizing the application using Docker.

Table of Contents

• Description

• Features

• Getting Started

  • Prerequisites

  • Installation

  • Running the Server

• Usage

  • API Endpoints

  • Examples

• Configuration

• License

Getting Started

Prerequisites

• Java: Java 11 or higher.

• Maven: Maven 3.6.0 or higher.

• Docker: Install Docker if you plan to use the Docker image.

Installation

1. Clone the repository:

   git clone [https://github.com/your-username/mcp-lucene-server.git](https://github.com/your-username/mcp-lucene-server.git)
   cd mcp-lucene-server
   

   (Replace your-username with your GitHub username)

2. Build the project using Maven:

   mvn clean install
   

Running the Server

Without Docker

1. Run the Spring Boot application:

   java -jar target/mcp-lucene-server-0.0.1-SNAPSHOT.jar

   (The exact name of the .jar file might vary slightly depending on your project version.)

2. The server will start on port 8080 by default.

With Docker

1. Ensure you have Docker installed: Follow the instructions on the official Docker website: https://docs.docker.com/get-docker/

2. Build the Docker image: Navigate to the root directory of your project in your terminal and run:

   docker build -t mcp-lucene-server .

3. Run the Docker container:

   docker run -p 8080:8080 mcp-lucene-server

   This will map port 8080 on your host machine to port 8080 inside the container.

MCP Shim for Claude Desktop

This project includes an optional MCP shim (mcp-shim/) that exposes the server's REST endpoints as MCP tools over STDIO so you can use them directly from Claude Desktop.

Prerequisites

• Java 17+
• Node.js 18+
• Maven 3.6+

1) Run the Spring Boot server

mvn spring-boot:run

The API will be available at http://localhost:8080/mcp/v1.

2) Run the MCP shim

cd mcp-shim
npm install
# JSON + text output (default)
LUCENE_BASE_URL=http://localhost:8080/mcp/v1 npm start
# If your client cannot render JSON tool outputs, force text-only
MCP_FORCE_TEXT=1 LUCENE_BASE_URL=http://localhost:8080/mcp/v1 npm start

3) Configure Claude Desktop

Update ~/.claude/mcp/config.json:

{
  "mcpServers": {
    "lucene": {
      "command": "/opt/homebrew/bin/node",
      "args": [".../MCP-Lucene-Server/mcp-shim/server.js"],
      "env": {
        "LUCENE_BASE_URL": "http://localhost:8080/mcp/v1",
        "MCP_FORCE_TEXT": "1"
      }
    }
  }
}

Alternatively, use the wrapper script to capture shim logs to /tmp/mcp-lucene-shim.stderr.log:

cat > .../MCP-Lucene-Server/mcp-shim/run-shim.sh <<'SH'
#!/usr/bin/env bash
set -euo pipefail
export LUCENE_BASE_URL="${LUCENE_BASE_URL:-http://localhost:8080/mcp/v1}"
exec node .../MCP-Lucene-Server/mcp-shim/server.js \
  2> /tmp/mcp-lucene-shim.stderr.log
SH
chmod +x .../MCP-Lucene-Server/mcp-shim/run-shim.sh

Then set in ~/.claude/mcp/config.json:

{
  "mcpServers": {
    "lucene": {
      "command": ".../MCP-Lucene-Server/mcp-shim/run-shim.sh",
      "env": {
        "LUCENE_BASE_URL": "http://localhost:8080/mcp/v1",
        "MCP_FORCE_TEXT": "1"
      }
    }
  }
}

4) Available tools

• lucene_status: Get server/index status
• lucene_upsert: Upsert documents
• lucene_query: Query documents (with optional metadata filters)
• lucene_delete: Delete by IDs
• lucene_list: List documents with pagination

5) Example prompts for Claude Desktop

• Run lucene_status
• Run lucene_list with: { "limit": 10, "offset": 0 }
• Run lucene_upsert with: {"documents":[{"id":"doc-1","text":"hello world","metadata":{"lang":"en"}}]}
• Run lucene_query with: {"queries":[{"query":"hello","top_k":5}]}
• Run lucene_delete with: { "ids": ["doc-1"] }

6) Troubleshooting

• Verify the API returns JSON:

curl -i http://localhost:8080/mcp/v1/status

• If Claude shows "unsupported format", start the shim with text-only output:

MCP_FORCE_TEXT=1 LUCENE_BASE_URL=http://localhost:8080/mcp/v1 npm start

• View shim logs (when using wrapper):

tail -n +1 /tmp/mcp-lucene-shim.stderr.log

• Ensure the paths in your config.json are absolute and correct, then restart Claude Desktop.

API Endpoints (for Curl)

The server provides the following API endpoints:

• GET /mcp/v1/status

  • Returns the status of the server.

• POST /mcp/v1/upsert

  • Upserts (inserts or updates) one or more documents.

  • Request body:

    {
      "documents": [
        {
          "id": "doc1",
          "text": "This is the text of document 1.",
          "metadata": {
            "category": "example",
            "language": "english"
          }
        },
        {
          "id": "doc2",
          "text": "This is document 2's text.",
          "metadata": {
            "category": "sample",
            "language": "spanish"
          }
        }
      ]
    }

• POST /mcp/v1/query

  • Queries the Lucene index.

  • Request body:

    {
      "queries": [
        {
          "query": "document",
          "top_k": 10,
          "filter": {
            "language": "english"
          }
        },
         {
          "query": "text search",
          "filter": {
             "category": "example"
           }
         }
      ]
    }

  • query: The Lucene query string.

  • top_k: (Optional) The maximum number of results to return (default: 10).

  • filter: (Optional) A map of metadata fields and values to filter by.

• POST /mcp/v1/delete

  • Deletes documents from the Lucene index.

  • Request body:

    {
        "ids": ["doc1", "doc2"]
    }

• GET /mcp/v1/list

  • Lists documents from the Lucene index.

  • Request body:

    {
        "ids": ["doc1", "doc2"]
    }

Examples

Get server status:

curl http://localhost:8080/mcp/v1/status

Upsert documents:

curl -X POST 

http://localhost:8080/mcp/v1/upsert 

-H 'Content-Type: application/json' 

-d '{
"documents": [
{
"id": "doc1",
"text": "This is the text of document 1.",
"metadata": {
"category": "example",
"language": "english"
}
},
{
"id": "doc2",
"text": "This is document 2''s text.",
"metadata": {
"category": "sample",
"language": "spanish"
}
}
]
}'

Query documents:

curl -X POST 

http://localhost:8080/mcp/v1/query 

-H 'Content-Type: application/json' 

-d '{
"queries": [
{
"query": "document text",
"top_k": 5,
"filter": {
"language": "english"
}
}
]
}'

Delete documents:

curl -X POST 

http://localhost:8080/mcp/v1/delete 

-H 'Content-Type: application/json' 

-d '{
"ids": ["doc1"]
}'

List documents:

curl -X POST 

http://localhost:8080/mcp/v1/list 

-H 'Content-Type: application/json' 

-d '{
"ids": ["doc1", "doc2"]
}'

Configuration

The server can be configured using Spring Boot's application properties. Here are some of the key properties:

• server.port: The port the server listens on (default: 8080).

• lucene.index.path: The path to the Lucene index directory. This is where the indexed data is stored. If not set, a default location is used. It is highly recommended to configure this to a persistent storage location.

You can set these properties in an application.properties or application.yml file in your src/main/resources directory, or by using environment variables.

Example application.properties:

server.port=8080 lucene.index.path=/path/to/lucene/index

License

This project is licensed under the Apache 2.0 License.