base120-mcp

120 validated mental models packaged as an MCP server to upgrade reasoning, prompt engineering, and architectural decisions in Claude Desktop, Cursor, and Windsurf.

Quickstart • The 6 Domains • Claude Desktop Setup • Engineering Verification

The 6 Reasoning Domains

Related MCP server: Tycana

What is Base120?

Most AI agents struggle with complex, ambiguous problems because they default to single-perspective reasoning.

Base120 is a structured library of 120 operational mental models (Inversion, First Principles, Ashby's Law of Requisite Variety, Nash Equilibrium, Antifragility) that give LLMs the exact operators needed to deconstruct hard engineering challenges.

Claude Desktop Setup

Add base120-mcp directly to your claude_desktop_config.json:

{
  "mcpServers": {
    "base120": {
      "command": "python",
      "args": ["-m", "base120_mcp.server"]
    }
  }
}

Now you can ask Claude in natural language:

"Use Base120 Inversion and Ashby's Law to critique our new distributed task queue architecture."

MCP Tools Provided

1. base120_search: Search mental models by domain or problem description.

2. base120_apply: Generate structured reasoning scaffolds for applying a specific model to an active problem statement.

Engineering Rigor & Verification

• Tests: 100% passing test suite across Python 3.10-3.13.

• Zero Dependencies: Pure Python standard library (json, sys, dataclasses).

• Author: Reuben Bowlby — LinkedIn | Resume

Enterprise Core

Base120 is the cognitive reasoning core of the HUMMBL platform. For enterprise multi-agent governance and formal Lean 4 verification, visit github.com/hummbl-io.