Boundary

Boundary MCP is a decision structure analysis engine built on the Model Context Protocol. It helps you translate vague decision anxiety into a structured analysis of your goals, constraints, uncertainties, and multiple strategy paths. You won’t get a prescriptive answer; instead you’ll see clear boundaries, option spaces, and cognitive blind spots so you can make your own informed choice.

How to use

You use Boundary MCP by connecting a compatible MCP client and describing your decision context in natural language. The system will translate your query into a decision structure, generate a set of strategy paths, and quantify risk distributions without claiming a single best outcome. You can explore multiple paths, compare their implications, and surface cognitive biases that might distort your judgment.

Typical usage flow:

• Describe a decision problem in your MCP client (for example, a career move, relocation, or major investment).
• Run analyze_decision to obtain a structured decision report containing goals, key variables, hard constraints, uncertainties, strategy space, and risk distributions.
• Review cognitive bias prompts to identify potential thinking errors and adjust your plan accordingly.
• Use the report to inform your personal decision, then proceed with your chosen path.

How to install

# Prerequisites
Node.js and npm must be installed on your machine.

# Quick start using npx (no installation required)
npx @eternalheart/mcp-boundary

# If you prefer a global install
npm install -g @eternalheart/mcp-boundary

# If you want to build from source (optional)
git clone https://github.com/wh131462/mcp-boundary.git
cd mcp-boundary
npm install
npm run build

# After building, you can run the local bundle if a start script is provided by the project
# Example (if a start script exists in the built bundle):
npm run start

Configuration and usage with an MCP client

You connect Boundary MCP to your MCP client using an MCP entry that runs the boundary tool. The simplest approach is to invoke Boundary MCP directly through its runtime command in your client configuration.

{
  "mcpServers": {
    "boundary": {
      "command": "npx",
      "args": ["@eternalheart/mcp-boundary"]
    }
  }
}

Notes on enabling with Claude Code (example integration)

You can add Boundary MCP as a server inside Claude Code by creating an MCP server entry named boundary that points to the runtime command.

claude mcp add boundary -- npx @eternalheart/mcp-boundary

Available tools

analyze_decision

Performs a complete decision structure analysis and returns a formatted report that includes problem restatement, variables and constraints, strategy space, risk distribution, and cognitive bias prompts.

get_analysis_json

Returns a structured JSON representation of the decision analysis for programmatic consumption.