ReasonKit

The AI Reasoning Engine

"From Prompt to Cognitive Engineering." — Turn Prompts into Protocols.

Auditable Reasoning for Production AI | Rust-Native | SSR/SSG Compatible

Website | Documentation | GitHub

---

The Problem We Solve

Most AI is a slot machine. Insert prompt → pull lever → hope for coherence.

ReasonKit is a factory. Input data → execute protocol → get deterministic, auditable output.

LLMs are fundamentally probabilistic. Same prompt → different outputs. This creates critical failures:

Failure	Impact	Our Solution
Inconsistency	Unreliable for production	Deterministic protocol execution
Hallucination	Dangerous falsehoods	Multi-source triangulation + adversarial critique
Opacity	No audit trail	Complete execution tracing with confidence scores

We don't eliminate probability (impossible). We constrain it through structured protocols that force probabilistic outputs into deterministic execution paths.

---

Quick Start

Prerequisites

• Rust toolchain installed (curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh)
• API key for your preferred LLM provider (e.g., OPENAI_API_KEY for OpenAI)
• Optional: uv for Python bindings (curl -LsSf https://astral.sh/uv/install.sh | sh)

# Install (Universal)
curl -fsSL https://reasonkit.sh/install | bash

# Or via Cargo
cargo install reasonkit-core

# Run your first analysis
rk-core think --profile balanced "Should we migrate to microservices?"

Note: In v0.1.x, CLI commands other than mcp, serve-mcp, and completions are scaffolded. The examples below describe the planned interface.

30 seconds to structured reasoning.

Troubleshooting

• Command not found: Ensure ~/.cargo/bin is in your PATH.
• API errors: Verify your API key is set and valid.
• Performance issues: Check system resources; ReasonKit targets <5ms latency.

---

ThinkTools: The 5-Step Reasoning Chain

Each ThinkTool acts as a variance reduction filter, transforming probabilistic outputs into increasingly deterministic reasoning paths.

ThinkTool	Operation	What It Does
GigaThink	Diverge()	Generate 10+ perspectives, explore widely
LaserLogic	Converge()	Detect fallacies, validate logic, find gaps
BedRock	Ground()	First principles decomposition, identify axioms
ProofGuard	Verify()	Multi-source triangulation, require 3+ sources
BrutalHonesty	Critique()	Adversarial red team, attack your own reasoning

Variance Reduction: The Chain Effect

Result: Raw LLM variance ~85% → Protocol-constrained variance ~28%

---

Reasoning Profiles

Pre-configured chains for different rigor levels:

# Fast analysis (70% confidence target)
rk-core think --profile quick "Is this email phishing?"

# Standard analysis (80% confidence target)
rk-core think --profile balanced "Should we use microservices?"

# Thorough analysis (85% confidence target)
rk-core think --profile deep "Design A/B test for feature X"

# Maximum rigor (95% confidence target)
rk-core think --profile paranoid "Validate cryptographic implementation"

Profile	Chain	Confidence	Use Case
--quick	GigaThink → LaserLogic	70%	Fast sanity checks
--balanced	All 5 ThinkTools	80%	Standard decisions
--deep	All 5 + meta-cognition	85%	Complex problems
--paranoid	All 5 + validation pass	95%	Critical decisions

Command Flow Diagram

flowchart TD
    A[User Input] --> B[Select Profile]
    B --> C[Execute ThinkTool Chain]
    C --> D[Output Result with Confidence]
    D --> E[Log Trace to SQLite]

Loading

Standard Operations:

# Balanced analysis (5-step protocol)
rk-core think --profile balanced "Should we migrate our monolith to microservices?"

# Quick sanity check (2-step protocol)
rk-core think --profile quick "Is this email a phishing attempt?"

# Maximum rigor (paranoid mode)
rk-core think --profile paranoid "Validate this cryptographic implementation"

# Scientific method (research & experiments)
rk-core think --profile scientific "Design A/B test for feature X"

With Memory (RAG):

# Ingest documents
rk-core ingest document.pdf

# Query with RAG
rk-core query "What are the key findings in the research papers?"

# View execution traces
rk-core trace list
rk-core trace export <id>

---

Reasoning Profiles

Pre-configured chains for different rigor levels:

# Fast analysis (70% confidence target)
rk-core think --profile quick "Is this email phishing?"

# Standard analysis (80% confidence target)
rk-core think --profile balanced "Should we use microservices?"

# Thorough analysis (85% confidence target)
rk-core think --profile deep "Design A/B test for feature X"

# Maximum rigor (95% confidence target)
rk-core think --profile paranoid "Validate cryptographic implementation"

Profile	Chain	Confidence	Use Case
--quick	GigaThink → LaserLogic	70%	Fast sanity checks
--balanced	All 5 ThinkTools	80%	Standard decisions
--deep	All 5 + meta-cognition	85%	Complex problems
--paranoid	All 5 + validation pass	95%	Critical decisions

---

See It In Action

$ rk-core think --profile balanced "Should we migrate to microservices?"

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
ThinkTool Chain: GigaThink → LaserLogic → BedRock → ProofGuard
Variance:        85% → 72% → 58% → 42% → 28%
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

[GigaThink] 10 PERSPECTIVES GENERATED                         Variance: 85%
  1. OPERATIONAL: Maintenance overhead +40% initially
  2. TEAM TOPOLOGY: Conway's Law - do we have the teams?
  3. COST ANALYSIS: Infrastructure scales non-linearly
  ...
  → Variance after exploration: 72% (-13%)

[LaserLogic] HIDDEN ASSUMPTIONS DETECTED                      Variance: 72%
  ⚠ Assuming network latency is negligible
  ⚠ Assuming team has distributed tracing expertise
  ⚠ Logical gap: No evidence microservices solve stated problem
  → Variance after validation: 58% (-14%)

[BedRock] FIRST PRINCIPLES DECOMPOSITION                      Variance: 58%
  • Axiom: Monoliths are simpler to reason about (empirical)
  • Axiom: Distributed systems introduce partitions (CAP theorem)
  • Gap: Cannot prove maintainability improvement without data
  → Variance after grounding: 42% (-16%)

[ProofGuard] TRIANGULATION RESULT                             Variance: 42%
  • 3/5 sources: Microservices increase complexity initially
  • 2/5 sources: Some teams report success
  • Confidence: 0.72 (MEDIUM) - Mixed evidence
  → Variance after verification: 28% (-14%)

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
VERDICT: conditional_yes | Confidence: 87% | Duration: 2.3s
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

What This Shows:

• Transparency: See exactly where confidence comes from
• Auditability: Every step logged and verifiable
• Deterministic Path: Same protocol → same execution flow
• Variance Reduction: Quantified uncertainty reduction at each stage

---

Architecture

The ReasonKit architecture uses a Protocol Engine wrapper to enforce deterministic execution over probabilistic LLM outputs.

Three-Layer Architecture:

1. Probabilistic LLM (Unavoidable)

   • LLMs generate tokens probabilistically
   • Same prompt → different outputs
   • We cannot eliminate this

2. Deterministic Protocol Engine (Our Innovation)

   • Wraps the probabilistic LLM layer
   • Enforces strict execution paths
   • Validates outputs against schemas
   • State machine ensures consistent flow

3. ThinkTool Chain (Variance Reduction)

   • Each ThinkTool reduces variance
   • Multi-stage validation catches errors
   • Confidence scoring quantifies uncertainty

Key Components:

• Protocol Engine: Orchestrates execution with strict state management
• ThinkTools: Modular cognitive operations with defined contracts
• LLM Integration: Unified client (Claude, GPT, Gemini, 18+ providers)
• Telemetry: Local SQLite for execution traces + variance metrics

Architecture (Mermaid Diagram)

flowchart LR
    subgraph CLI["ReasonKit CLI (rk-core)"]
      A[User Command<br/>rk-core think --profile balanced]
    end

    subgraph PROTOCOL["Deterministic Protocol Engine"]
      B1[State Machine<br/>Execution Plan]
      B2[ThinkTool Orchestrator]
      B3[(SQLite Trace DB)]
    end

    subgraph LLM["LLM Layer (Probabilistic)"]
      C1[Provider Router]
      C2[Claude / GPT / Gemini / ...]
    end

    subgraph TOOLS["ThinkTools · Variance Reduction"]
      G["GigaThink<br/>Diverge()"]
      LZ["LaserLogic<br/>Converge()"]
      BR["BedRock<br/>Ground()"]
      PG["ProofGuard<br/>Verify()"]
      BH["BrutalHonesty<br/>Critique()"]
    end

    A --> B1 --> B2 --> G --> LZ --> BR --> PG --> BH --> B3
    B2 --> C1 --> C2 --> B2

    classDef core fill:#030508,stroke:#06b6d4,stroke-width:1px,color:#f9fafb;
    classDef tool fill:#0a0d14,stroke:#10b981,stroke-width:1px,color:#f9fafb;
    classDef llm fill:#111827,stroke:#a855f7,stroke-width:1px,color:#f9fafb;

    class CLI,PROTOCOL core;
    class G,LZ,BR,PG,BH tool;
    class LLM,llm C1,C2;

Loading

---

Built for Production

ReasonKit is written in Rust because reasoning infrastructure demands reliability.

Capability	What It Means for You
Predictable Latency	<5ms orchestration overhead, no GC pauses
Memory Safety	Zero crashes from null pointers or buffer overflows
Single Binary	Deploy anywhere, no Python environment required
Fearless Concurrency	Run 100+ reasoning chains in parallel safely
Type Safety	Errors caught at compile time, not runtime

Benchmarked Performance (view full report):

Operation	Time	Target
Protocol orchestration	4.4ms	<10ms
RRF Fusion (100 elements)	33μs	<5ms
Document chunking (10 KB)	27μs	<5ms
RAPTOR tree traversal (1000 nodes)	33μs	<5ms

Why This Matters:

Your AI reasoning shouldn't crash in production. It shouldn't pause for garbage collection during critical decisions. It shouldn't require complex environment management to deploy.

ReasonKit's Rust foundation ensures deterministic, auditable execution every time—the same engineering choice trusted by Linux, Cloudflare, Discord, and AWS for their most critical infrastructure.

---

Memory Infrastructure (Optional)

Memory modules (storage, embedding, retrieval, RAPTOR, indexing) are available in the standalone reasonkit-mem crate.

Enable the memory feature to use these modules:

[dependencies]
reasonkit-core = { version = "0.1", features = ["memory"] }

Features:

• Qdrant vector database (embedded mode)
• Hybrid search (dense + sparse fusion)
• RAPTOR hierarchical retrieval
• Local embeddings (BGE-M3 ONNX)
• BM25 full-text search (Tantivy)

---

Installation

Primary Method (Universal):

curl -fsSL https://reasonkit.sh/install | bash

Alternative Methods

# Cargo (Rust) - Recommended for Developers
cargo install reasonkit-core

# From Source (Latest Features)
git clone https://github.com/reasonkit/reasonkit-core
cd reasonkit-core && cargo build --release

Python bindings available via PyO3 (build from source with --features python).

---

Usage Examples

Pre-configured chains for different rigor levels:

# Fast analysis (70% confidence target)
rk-core think --profile quick "Is this email phishing?"

# Standard analysis (80% confidence target)
rk-core think --profile balanced "Should we use microservices?"

# Thorough analysis (85% confidence target)
rk-core think --profile deep "Design A/B test for feature X"

# Maximum rigor (95% confidence target)
rk-core think --profile paranoid "Validate cryptographic implementation"

Profile	Chain	Confidence	Use Case
--quick	GigaThink → LaserLogic	70%	Fast sanity checks
--balanced	All 5 ThinkTools	80%	Standard decisions
--deep	All 5 + meta-cognition	85%	Complex problems
--paranoid	All 5 + validation pass	95%	Critical decisions

Command Flow Diagram

flowchart TD
    A[User Input] --> B[Select Profile]
    B --> C[Execute ThinkTool Chain]
    C --> D[Output Result with Confidence]
    D --> E[Log Trace to SQLite]

Loading

Standard Operations:

# Balanced analysis (5-step protocol)
rk-core think --profile balanced "Should we migrate our monolith to microservices?"

# Quick sanity check (2-step protocol)
rk-core think --profile quick "Is this email a phishing attempt?"

# Maximum rigor (paranoid mode)
rk-core think --profile paranoid "Validate this cryptographic implementation"

# Scientific method (research & experiments)
rk-core think --profile scientific "Design A/B test for feature X"

With Memory (RAG):

# Ingest documents
rk-core ingest document.pdf

# Query with RAG
rk-core query "What are the key findings in the research papers?"

# View execution traces
rk-core trace list
rk-core trace export <id>

---

Contributing: The 5 Gates of Quality

We demand excellence. All contributions must pass The 5 Gates of Quality:

# Clone & Setup
git clone https://github.com/reasonkit/reasonkit-core
cd reasonkit-core

# The 5 Gates (MANDATORY)
cargo build --release        # Gate 1: Compilation (Exit 0)
cargo clippy -- -D warnings  # Gate 2: Linting (0 errors)
cargo fmt --check            # Gate 3: Formatting (Pass)
cargo test --all-features    # Gate 4: Testing (100% pass)
cargo bench                  # Gate 5: Performance (<5% regression)

Quality Score Target: 8.0/10 minimum for release.

See CONTRIBUTING.md for complete guidelines.

---

🏷️ Community Badge

If you use ReasonKit in your project, add our badge:

[![Reasoned By ReasonKit](https://raw.githubusercontent.com/reasonkit/reasonkit-core/main/brand/badges/reasoned-by.svg)](https://reasonkit.sh)

See Community Badges for all variants and usage guidelines.

---

🎨 Branding & Design

• Brand Playbook - Complete brand guidelines
• Component Spec - UI component system
• Motion Guidelines - Animation system
• 3D Assets - WebGL integration guide
• Integration Guide - Complete integration instructions

---

Design Philosophy: Honest Engineering

We don't claim to eliminate probability. That's impossible. LLMs are probabilistic by design.

We do claim to constrain it. Through structured protocols, multi-stage validation, and deterministic execution paths, we transform probabilistic token generation into auditable reasoning chains.

What We Battle	How We Battle It	What We're Honest About
Inconsistency	Deterministic protocol execution	LLM outputs still vary, but execution paths don't
Hallucination	Multi-source triangulation, adversarial critique	Can't eliminate, but can detect and flag
Opacity	Full execution tracing, confidence scoring	Transparency doesn't guarantee correctness
Uncertainty	Explicit confidence metrics, variance reduction	We quantify uncertainty, not eliminate it

---

Version & Maturity

Component	Status	Notes
ThinkTools Chain	✅ Stable	Core reasoning protocols production-ready
MCP Server	✅ Stable	Model Context Protocol integration
CLI	🔶 Scaffolded	mcp, serve-mcp, completions work; others planned
Memory Features	✅ Stable	Via reasonkit-mem crate
Python Bindings	🔶 Beta	Build from source with --features python

Current Version: v0.1.2 | CHANGELOG | Releases

Verify Installation

# Check version
rk-core --version

# Verify MCP server starts
rk-core serve-mcp --help

# Run a quick test (requires LLM API key)
OPENAI_API_KEY=your-key rk-core mcp

---

License

Apache 2.0 - See LICENSE

Open Source Core: All core reasoning protocols and ThinkTools are open source under Apache 2.0.

---

ReasonKit — Turn Prompts into Protocols

Designed, Not Dreamed

Website | Documentation | GitHub