conductor-tasks

Conductor Tasks is an AI-powered task management assistant designed for AI development workflows. It can parse PRDs, Markdown, and notes into structured tasks, expand those tasks into detailed implementation plans, and provide AI-assisted code guidance. It supports a multi-LLM setup, allowing you to route different tasks to OpenAI, Anthropic, Groq, Mistral, Google Gemini, Perplexity, xAI, Azure OpenAI, or local endpoints, all from a single interface. Through MCP integration, you can interact with Conductor Tasks directly inside your editor or use its standalone CLI to manage tasks, visualize Kanban boards, dependency trees, and dashboards, and generate or modify code changes using AI assistance.

To use it via MCP, configure the server in your editor's MCP settings as shown in the Quick Start example. You can then chat with the AI to initialize projects, parse PRDs, list tasks, fetch next actions, generate detailed implementation steps, or visualize task structures. The CLI offers equivalent commands for project initialization, PRD parsing, task listing, and task updates, with specialized commands like generate-steps, visualize, and get-details to drive the workflow.

Key capabilities include: AI-powered task generation from PRDs or notes, expansion into step-by-step implementation plans, AI-assisted code modification and diffs, and built-in visualizations (kanban boards, dependency trees, dashboards). You can also template and template-driven task creation to standardize workflows across projects, making it a flexible co-pilot for the entire development lifecycle.

Prerequisites:

• Node.js and npm installed on your system
• Access to an environment where you can install global npm packages or run npx
• API keys for your preferred LLM providers (e.g., OpenAI) and any other integrations you plan to use

Install options:

Option A: Install globally for CLI use

# Install globally (recommended for CLI use)
npm install -g conductor-tasks

# Or use npx without installing globally
# npx conductor-tasks <command>

Option B: Use via npx without installation

# Run the MCP-enabled server via npx
npx conductor-tasks --serve-mcp

Environment setup:

# Create a .env file or export variables as needed
export OPENAI_API_KEY=your-openai-key
export ANTHROPIC_API_KEY=your-anthropic-key
export GOOGLE_API_KEY=your-google-key
# Set the default provider or per-task overrides as needed
export DEFAULT_LLM_PROVIDER=openai

Configuration in editor (MCP):

• Add the MCP server configuration as shown in the README example under mcpServers.

For detailed usage, refer to the documentation in the docs directory and the CLI help.

Tips and notes:

• Ensure you provide valid API keys for the LLM providers you intend to use. The DEFAULT_LLM_PROVIDER can be overridden per task or per session as needed.
• When using visualizations, you may need additional dependencies or browser support depending on your editor or CLI environment.
• If you encounter rate limits, consider routing specific tasks to alternative providers or enabling local endpoints like Ollama or LM Studio for offline/local processing.
• Use the recommendable commands like parse-prd, generate-steps, and visualize to streamline task planning and execution.
• Keep your PRD and template files up to date to maximize AI accuracy in task generation and expansion.