Quickstart

Connect your model to the reasoning engine

One JSON block. Your model keeps its own API key — the Kohnex Brain engine adds expert reasoning on top of every answer.

{
  "mcpServers": {
    "kohnex": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://kohnex.com/sse"],
      "env": { "API_KEY": "khnx_..." }
    }
  }
}

Core concept

The Kohnex Brain engine

Kohnex is a reasoning engine, not a library. It holds expert reasoning inbuilt — a 1,800-node knowledge graph across 38+ domains — and it creates new reasoning on demand when nothing inbuilt fits.

Reference

Engine tools

Every tool calls the same engine. Always start with maple() — it finds the reasoning your question needs.

Find reasoning

Load reasoning

Combine reasoning

Create reasoning

Knowledge engine

Reasoning methodologies

Reference

Workflow

The standard flow: the engine finds the reasoning, loads it, your model follows it.

User asks question
  → maple()          Engine finds the right reasoning
  → Pick top         From scored candidates
  → cedar(id)        Engine loads the full steps
  → Follow steps     Model reasons like an expert
  → Answer           Structured, expert-level

Reference

Reasoning domains (22)

The domains the engine reasons across. maple() routes your question to the right one.

Reference

Examples

Two sample prompts showing the engine flow.

Ask the engine (maple) for security reasoning → load it (cedar) → answer like a threat analyst

Ask the engine (maple) for performance reasoning → load it (cedar) → answer like a staff engineer