Error Tracker

You run an Error Tracker MCP Server that automates code error handling by tracing the code owner, gathering pull request details, and creating JIRA tasks. This server is designed to work with MCP-enabled clients to streamline error investigation and issue creation across your software projects.

How to use

You connect a client to the Error Tracker MCP Server to perform end-to-end error investigations. Typical workflows include finding the code owner for a given file and line, locating the related pull request, automatically investigating the error, and creating a JIRA ticket based on the gathered information. For remote deployments, you use an MCP URL; for local development, you run the server locally and connect via the stdio interface.

How to install

Prerequisites: you need Node.js installed on your system. Ensure you have a modern Node.js runtime to build and run the server.

Step 1: Install dependencies.

npm install

Step 2: Build the server.

npm run build

Step 3: Prepare environment variables. Create a file named .env and populate the required values as shown in the example below.

# 服务器配置
HTTP_PORT=3000                    # HTTP 模式端口（默认 3000）
COMPATIBLE_PORT=3001              # 兼容模式端口（默认 3001）
DEFAULT_BRANCH=main               # 默认调查分支（默认 main）

# Bitbucket 配置
BITBUCKET_USERNAME=your_username
BITBUCKET_PASSWORD=your_password
BITBUCKET_BASE_URL=https://your-bitbucket-server.com
BITBUCKET_PROJECT=your_project
BITBUCKET_REPO=your_repo
PR_TARGET_BRANCH=main             # PR 目标分支过滤（默认 main）

# JIRA 配置
JIRA_USERNAME=your_username
JIRA_PASSWORD=your_password
JIRA_BASE_URL=https://your-jira-server.com
JIRA_PROJECT_KEY=YOUR_PROJECT
JIRA_ISSUE_TYPE_ID=10101
JIRA_PRIORITY_ID=10000
JIRA_COMPONENT_ID=12505

How to run in different modes

There are three modes you can use to run the server. Choose the mode that matches your deployment scenario and client compatibility.

• stdio mode for local use (Claude Desktop) runs the server within your local environment.

• HTTP mode for remote deployment uses the MCP HTTP endpoint.

• compatible mode supports both new and old MCP clients in remote deployments.

Additional sections

Configuration you will use in MCP clients is shown below. The stdio configuration runs locally, while the HTTP configuration connects remotely.

Client configuration (stdio)

{
  "mcpServers": {
    "error-tracker": {
      "command": "node",
      "args": ["-r", "dotenv/config", "/absolute/path/error/build/index.js"]
    }
  }
}

Client configuration (http)

Choose the HTTP endpoint if you are using a remote MCP client. The Streamable HTTP endpoint is the recommended option for modern MCP clients.

{
  "mcpServers": {
    "error-tracker": {
      "url": "http://localhost:3000/mcp"
    }
  }
}

Client configuration (compatibility mode)

If your client only supports the older MCP protocol, use the compatibility endpoint.

{
  "mcpServers": {
    "error-tracker": {
      "url": "http://localhost:3000/sse"
    }
  }
}

Remote deployment and docker

You can deploy the server remotely using Docker. Build and run the container with the environment file to apply all required settings.

FROM node:18-alpine
WORKDIR /app
COPY package*.json ./
RUN npm ci --only=production
COPY . .
RUN npm run build
EXPOSE 3000
CMD ["node", "-r", "dotenv/config", "build/compatible-server.js"]

Build and run the image with the environment file.

docker build -t error-tracker-mcp .
docker run -d -p 3000:3000 --env-file .env error-tracker-mcp

Process management with PM2

You can manage the server with PM2 for production workflows.

pm2 start npm --name "error-tracker" -- run start:compatible
pm2 save && pm2 startup

Troubleshooting

Common issues and fixes help you keep the server running smoothly.

• Authentication failures: verify your .env file contains correct credentials for Bitbucket and JIRA.

• Missing files: ensure the file path is correct relative to the repository root and confirm DEFAULT_BRANCH is set as intended.

• Port conflicts: adjust HTTP_PORT or COMPATIBLE_PORT in the .env file.

Technical notes

Tech stack includes TypeScript, MCP SDK, Express, and Axios. The server supports stdio, Streamable HTTP, and a compatibility mode to accommodate different client versions.

License

MIT license.

Available tools

find_code_owner

Identify the code owner by file path and line number to determine responsibility for a specific code segment.

get_pull_request

Locate the related pull request based on a commit ID to gather context and discussion around the change.

investigate_error

Automatically investigate an error by combining ownership data and PR details to prepare for ticket creation.

create_jira_ticket

Create and assign a Jira ticket using the findings from the investigation, including a summary and labels.

track_error_full

Complete end-to-end process: investigate, analyze, and create a Jira ticket in one streamlined action.