mcp-server by bitDive - MCP Server

BitDive MCP Server

A Spring Boot application acting as a Model Context Protocol (MCP) server. It provides AI agents with tools to interact with the BitDive monitoring system, enabling retrieval of trace data, service maps, and error analysis.

🚀 Features

This server exposes tools that allow AI assistants to query monitoring data directly:

Trace Tools

findTraceAll: Returns the full call trace for a specified Call ID.
findTraceForMethod: Retrieves the trace for a specific method within a given Call ID.
findTraceForMethodBetweenTime: Searches for specific method executions within a defined time range.

Monitoring & Performance Tools

getCurrentHeapMapAllSystem: Returns system performance metrics for the entire system (Heat Map).
getCurrentHeapMapForModule: Returns performance metrics filtered by a specific module.
getCurrentHeapMapForModuleAndForService: Returns performance metrics for a specific service within a module.
getCurrentHeapMapForModuleAndForServiceClass: Returns performance metrics drilling down to a specific class within a service.
getLastCallService: Retrieves a list of recent execution traces (calls) for a specific service.

🛠 Technology Stack

Java: 17
Framework: Spring Boot 3.2.0
AI Integration: Spring AI 1.0.0 (MCP Server WebFlux)
Database: PostgreSQL
Security: HashiCorp Vault, Bouncy Castle

⚙️ Configuration

The application runs on port 8089 by default. Configuration is managed via application.yml and can be overridden with environment variables:

Variable	Description	Default
`POSTGRES_URL`	JDBC URL	`jdbc:postgresql://37.27.0.220:5432/data-bitdive`
`POSTGRES_USER`	DB Username	`citizix_user`
`POSTGRES_PASS`	DB Password	`S3cret`
`VAULT_URL`	Vault URL	`https://sandbox.bitdive.io/vault`
`TOKEN_SECRET`	Token Secret	(See application.yml)

📦 How to Run

Prerequisites

JDK 17 or higher
Maven (wrapper provided)

Build & Run

Build the project:
```
./mvnw clean package
```
Start the server:
```
./mvnw spring-boot:run
```

MCP Connection

Once running, the server exposes the following endpoints for MCP clients:

SSE Endpoint: http://localhost:8089/sse
Message Chat Endpoint: http://localhost:8089/mcp/message

To use this with an MCP Client (like Cursor), add it to your configuration as an SSE server.

🐳 Deployment

The pom.xml is configured to copy the resulting JAR file to ../docker/docker-file-mcp-server during the package phase, facilitating Docker builds.

🔌 Client Integration (Cursor / Claude)

To connect your AI tool to the BitDive MCP Server, follow these steps:

1. Configure the MCP Server

Add the following configuration to your mcp.json (usually located in .cursor/mcp.json or similar).

Example: Connection to BitDive Cloud (SaaS)

{
  "mcpServers": {
    "bitdive": {
      "url": "https://cloud.bitdive.io/mcp/sse",
      "name": "BitDive MCP Server"
    }
  }
}

For Self-Hosted Infrastructure: If you have deployed the BitDive Infrastructure on your own server, replace https://cloud.bitdive.io with your custom domain (e.g., https://your-domain.com/mcp/sse).

2. Authentication (Temporary)

Current limitations require passing the API key via prompts/rules. A seamless "One-Click Integration" is coming in the next sprint.

Deploy Infrastructure: You need a running instance of the BitDive platform.
- 👉 Get started with BitDive Infrastructure
Get your Token: Log in to your BitDive dashboard and navigate to the MCP Integration section to copy your API Key.
Add User Rule: In your AI tool (e.g., Cursor Settings > General > Rules for AI), add the following line:
```
for API key for mcp bitDive access use <YOUR_TOKEN_HERE>
```