Backlog Manager

Backlog Manager is an MCP (Machine-Consumable Programming) server for issue and task management with a file-based approach. It provides tools for AI agents and other clients to create issues, add tasks to them, and track task status. Issues represent high-level feature requests or bugs, while tasks represent specific work items needed to resolve the issue.

Built using Anthropic's MCP protocol, it supports both SSE and stdio transports for flexible integration with AI assistants like Claude, or other MCP-compatible clients.

- Issue Management: Create, list, select, and track issues with descriptions
- Task Tracking: Add tasks to issues with titles, descriptions, and status tracking
- Status Workflow: Track task progress through New, InWork, and Done states
- File-Based Storage: Portable JSON storage format for easy backup and version control
- Flexible Transport: Support for both SSE (HTTP) and stdio communication
- Docker Support: Run in containers for easy deployment and isolation

- Python: 3.12 or higher
- Package Manager: uv (recommended) or pip
- Docker: (Optional) For containerized deployment
- MCP Client: Claude Code, Windsurf, or any other MCP-compatible client

# Clone the repository git clone https://github.com/username/backlog-manager-mcp.git cd backlog-manager-mcp # Install dependencies uv pip install -e . # Verify installation uv run backlog-manager # This should start the server

# Build the Docker image docker build -t backlog/manager --build-arg PORT=8050 . # Run the container docker run -p 8050:8050 backlog/manager # Verify container is running docker ps | grep backlog/manager

Configure the server behavior using environment variables in a.envfile:

# Create environment file from example cp .env.example .env

# Transport mode: 'sse' or 'stdio' TRANSPORT=sse # Server configuration (for SSE transport) HOST=0.0.0.0 PORT=8050 # Data storage TASKS_FILE=tasks.json

# Using the CLI command uv run backlog-manager # Or directly with Python uv run src/backlog_manager/main.py

The Backlog Manager exposes the following tools via MCP:

Tasks and issues can have one of the following statuses:

- New(default for new tasks/issues)
- InWork(in progress)
- Done(completed)

Once you have the server running with SSE transport, connect to it using this configuration:

{ "mcpServers": { "backlog-manager": { "transport": "sse", "url": "http://localhost:8050/sse" } } }

{ "mcpServers": { "backlog-manager": { "transport": "sse", "serverUrl": "http://localhost:8050/sse" } } }

Usehost.docker.internalinstead oflocalhostto access the host machine from n8n container: