ReasonKit

The Reasoning Engine — Auditable Reasoning for Production AI

Website • Documentation • Crates.io

---

Overview

ReasonKit transforms ad-hoc LLM prompting into structured, auditable reasoning chains.

Philosophy: Designed, Not Dreamed — Structure beats raw intelligence.

Installation

# Install complete suite
cargo install reasonkit

# Or install individual components
cargo install reasonkit-core  # Reasoning engine
cargo install reasonkit-mem   # Memory layer
cargo install reasonkit-web   # Web sensing

Quick Start

# Run structured reasoning
reasonkit think --profile balanced "Should we adopt microservices?"

# Search knowledge base
reasonkit mem search "machine learning fundamentals"

# Start MCP server
reasonkit serve

Components

Crate	Description
reasonkit	Meta-crate — installs complete suite
reasonkit-core	The Reasoning Engine — ThinkTools, protocols, MCP
reasonkit-mem	Memory layer — vector storage, hybrid search, RAPTOR
reasonkit-web	Web sensing — browser automation, content extraction

Architecture

reasonkit (meta-crate)
│
├── reasonkit-core    ─── The Reasoning Engine
│   ├── ThinkTools (GigaThink, LaserLogic, BedRock, ProofGuard, BrutalHonesty)
│   ├── Protocol Executor
│   ├── LLM Client (18+ providers)
│   └── MCP Server
│
├── reasonkit-mem     ─── Memory Infrastructure
│   ├── Vector Storage (Qdrant)
│   ├── Sparse Index (Tantivy BM25)
│   ├── Hybrid Retrieval
│   └── RAPTOR Trees
│
└── reasonkit-web     ─── Web Sensing Layer
    ├── Browser Controller (CDP)
    ├── Content Extraction
    └── MCP Sidecar

License

Apache 2.0 — See LICENSE for details.

---

reasonkit.sh — Turn Prompts into Protocols