HUMMBL MCP Server

How to use

You interact with the MCP server through a client that can send requests for models, transformations, problem patterns, and guided workflows. You can retrieve detailed information about a specific model, list all models by transformation, search models by keywords, get AI-driven recommendations for complex problems, and step through guided problem‑solving workflows. Use these capabilities to explore, organize, and apply mental models to real-world challenges.

How to install

Prerequisites: ensure you have a modern Node.js runtime and npm installed on your system.

# Option A: Global installation (recommended)
npm install -g @hummbl/mcp-server

# Option B: Run using NPX without installation
npx @hummbl/mcp-server

Configuration and usage notes

Configure your Claude Desktop integration to access the HUMMBL MCP Server from within your workflow environment. Create or update your Claude Desktop configuration with an MCP server entry that runs the MCP server via NPX. This will expose the HUMMBL tools in the attachment menu for quick access.

{
  "mcpServers": {
    "hummbl": {
      "command": "npx",
      "args": ["-y", "@hummbl/mcp-server"]
    }
  }
}

Examples of tool interactions

The server exposes a set of tools you can invoke to explore models, transformations, and problem patterns. For example, you can fetch a specific model by code, list all models under a transformation, or search for models related to a keyword. You can also request recommendations for a complex scenario and retrieve transformation details.

Guided workflows and problem solving

Guided workflows provide structured, multi-step problem solving. You can list available workflows, start a workflow for a given problem description, and proceed through each step as you gather insights. Each workflow defines a sequence of transformations to apply (for example, Root Cause Analysis uses Perspective → Inversion → Decomposition → Meta-Systems).

Available resources and endpoints

Access models and transformations directly through the URIs and endpoints exposed by the MCP server. You can retrieve individual models, transformation sets, the entire Base120 framework, and methodologies.

Problem patterns and methodology

HUMMBL includes predefined problem patterns that map common challenges to recommended transformations and models. You can apply the Self-Dialectical AI methodology, which combines five dialectical stages with Base120 models and the SY meta-models to guide self-correction and rigorous analysis.

Development and maintenance

Initial development follows a standard Node.js project structure. Build and run locally to verify changes before deploying to production.

Notes on security and usage

Protect any API keys or credentials and rotate them as needed. When integrating with external tools, ensure only approved endpoints and configurations are used.

get_model

Retrieve detailed information about a specific mental model by its code (e.g., P1).

list_all_models

List all 120 mental models, optionally filtered by transformation type (e.g., P).

search_models

Search mental models by keyword across names, descriptions, and examples.

recommend_models

Get AI-recommended models based on a problem description.

get_transformation

Retrieve information about a specific transformation type and all its models.

search_problem_patterns

Find predefined problem patterns with recommended approaches.

list_workflows

List all available guided workflows.

start_workflow

Begin a guided workflow for your problem with a description.

continue_workflow

Proceed to the next step after completing the current one.

find_workflow_for_problem

Discover which workflow best fits your problem using keywords.