A hierarchical planning and task management tool with powerful MCP server integration for AI assistants

Scatterbrain keeps an AI Agent on task, so you don't have to. It does this by treating the AI to act as a planner and implementer, in alternating fashion. This helps the AI break down complex problems into manageable, hierarchical tasks and more importantly, keeps it on track. Its primary strength lies in its Model Context Protocol (MCP) server, which seamlessly integrates with AI assistants like Claude and Cursor, enabling intelligent task management and planning workflows.

curl --proto '=https' --tlsv1.2 -LsSf https://github.com/owen-d/scatterbrain/releases/latest/download/scatterbrain-installer.sh | sh

powershell -ExecutionPolicy Bypass -c "irm https://github.com/owen-d/scatterbrain/releases/latest/download/scatterbrain-installer.ps1 | iex"

# Install Rust if you haven't already
curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh

# Clone and build scatterbrain
git clone https://github.com/owen-d/scatterbrain.git
cd scatterbrain
cargo build --release

# Install globally (optional)
cargo install --path .

Once installed, all you need to do is add the following to your Cursor MCP configuration (expose is optional, but allows you to follow progress via the web interface):

{
  "mcpServers": {
    "scatterbrain": {
      "command": "scatterbrain",
      "args": [
        "mcp",
        "--expose",
        "3099"
      ],
      "env": {}
    }
  }
}

Scatterbrain's hierarchical planning system helps you break down complex projects into manageable tasks across multiple abstraction levels:

As you work through your plan, the interface shows your progress and maintains context across all levels:

• "leveled" abstraction system: Planning → Isolation → Ordering → Implementation
• Break complex projects into manageable pieces
• Natural progression from high-level goals to specific actions

• specialized tools for AI assistants
• Plan creation, task management, navigation, and notes
• Seamless integration with Cursor, Claude, and other MCP-compatible tools

• Modern browser-based UI for visual task management
• Real-time updates and intuitive navigation
• Access via --expose flag when running MCP server

• Complete command-line interface for all operations
• Shell completions for bash, zsh, fish
• Environment variable support for workflow integration

scatterbrain mcp --expose 8080

# Create a plan
scatterbrain plan create "Build documentation system"

# Add tasks and navigate
export SCATTERBRAIN_PLAN_ID=1
scatterbrain task add --level 0 "Design architecture" --notes "High-level planning"
scatterbrain move 0

scatterbrain serve --port 3000

scatterbrain mcp --expose 8080

Scatterbrain organizes work across four abstraction levels:

1. Planning (Level 0) - High-level goals and architecture
2. Isolation (Level 1) - Independent components and boundaries
3. Ordering (Level 2) - Sequence and dependencies
4. Implementation (Level 3) - Concrete, actionable tasks

When running as an MCP server, scatterbrain provides these tools to AI assistants:

• get_plan - Retrieve plan details and structure
• create_plan - Create new plans from prompts
• add_task - Add tasks at appropriate abstraction levels
• complete_task - Mark tasks as completed with summaries
• move_to - Navigate between tasks
• get_current - Get current task context
• get_distilled_context - Get focused planning context
• And 10 more specialized tools for comprehensive task management

This project is licensed under the MIT License - see the LICENSE file for details.