mcp-task-orchestrator

The MCP Task Orchestrator is a Python-based MCP server that automatically documents decisions, implementations, and tests as you develop. It provides specialist AI roles (Architect, Implementer, Tester, Documenter, and more) and persistent memory to keep project context intact across interactions. The server analyzes your project structure, saves artifacts in an organized way, and exposes a REST-like workflow through your MCP clients so you can guide a project from design to implementation with clearly defined steps.

To use it, install the package and run the server via the Python module entry point. Once running, configure your MCP client to connect to this server and begin sessions. You can prompt the orchestrator to initialize a session, plan tasks, and allocate roles for each subtask. The system will orchestrate the steps, provide structured workflows, and generate artifacts such as design documents, code scaffolds, tests, and troubleshooting guides as you progress.

Prerequisites

• Python 3.8+ (as indicated by the project)
• An MCP client (Claude Desktop, Cursor IDE, Windsurf, VS Code with extensions, etc.)

Installation steps

1. Install the MCP Task Orchestrator from PyPI

pip install mcp-task-orchestrator

2. Configure your MCP client to connect to the orchestrator (see client documentation for how to point at the server). If you are running locally, the server will be available after you run the next step.

3. Run the orchestrator module

python -m mcp_task_orchestrator

4. Optional: install with development or specific client support

python -m mcp_task_orchestrator --dev
python -m mcp_task_orchestrator --clients claude,cursor

5. Verify installation and status

python -m mcp_task_orchestrator --help
python -m mcp_task_orchestrator --status

Additional installation notes:

• For externally managed environments (WSL, Ubuntu, etc.), consider using a virtual environment:

python -m venv mcp-env
source mcp-env/bin/activate
pip install mcp-task-orchestrator

• You can also install via pipx for isolation (if desired):

pipx install mcp-task-orchestrator

Tips and gotchas:

• Ensure you have Python 3.8 or newer and a compatible MCP client installed before running the orchestrator.
• The orchestrator saves artifacts to a project-root-aware storage location; ensure your working directory is within or tied to your project root for proper artifact placement.
• If you run into issues with client detection, verify that the MCP client CLI tools (e.g., claude, cursor) are installed and accessible in your PATH.
• Use the --help and --status options frequently to confirm the server state and available options.
• You can customize roles and workflows via configuration files in your project (e.g., roles.yaml) to tailor the specialist AI behavior to your project.
• For development workflows, the --dev flag installs tools and dependencies useful for contributors but may pull in extra packages; use --venv or --user to isolate installations as needed.