Early Access • Written in Rust • Apache 2.0 • GitHub →

From Prompt Engineering to Cognitive Engineering

Tree-of-Thoughts achieves 74% success where Chain-of-Thought achieves 4%.
ReasonKit brings systematic divergent thinking to Claude, Gemini, and your AI agents.

"Should I take this job offer?"
Typical AI 😐
Consider salary, benefits, and culture fit.
With ReasonKit 🎯
Your manager = 80% of job satisfaction. Who will it be? Nobody's answering that.
curl -fsSL https://get.reasonkit.sh | bash
cargo install reasonkit
12,400+ installs • ★ 2.1K GitHub • < 100ms avg response
WORKS WITH
Claude Gemini OpenAI Cursor VS Code Any LLM
reasonkit-core
⏱ 89ms 94% confidence
GigaThink
LaserLogic
BedRock
ProofGuard
BrutalHonesty

Claude Code

Wrap your CLI sessions with structured reasoning. Use reasonkit wrap claude to force logical consistency.

Cursor & VS Code

IDE extension for inline logic checks. Highlight code and ask: "Is this logic sound?" directly in your editor.

Model Context Protocol

Expose reasoning as an MCP server. Let any LLM access ReasonKit's structured analysis tools.

Python SDK

Add critical thinking to your LangChain agents. import reasonkit to validate agent outputs.

CI/CD Pipelines

Block PRs that fail logic gates. Automated reasoning checks for high-stakes codebases.

Deep Research

Autonomous web triangulation. Verify claims against 3+ independent sources automatically.

The Problem

AI Gives You Answers. But Are They Good?

Most AI responses sound helpful but miss the hard questions.

You ask:
"Should I take this job offer?"
"Consider salary, benefits, and culture fit."
What's missing:
Manager quality, team turnover, company trajectory, your leverage, opportunity cost, where people go after 2-3 years...
You ask:
"Should I buy a house?"
"Building equity is generally better long-term."
What's missing:
Mortgage interest isn't equity. Transaction costs. Maintenance. Flexibility. The actual math shows breakeven at 5-7 years.
You ask:
"I want to start a YouTube channel"
"Focus on a niche and be consistent!"
What's missing:
Median income is $0. 10-40 hours per video. 95% quit in year one. MrBeast has 100 people working for him.
ThinkTools

Five Tools. Five Blind Spots. Zero BS.

Each ThinkTool catches a specific type of oversight.

GigaThink

See All The Angles
The blind spot: You're only thinking about a decision from one angle.
Without GigaThink
TYPICAL AI RESPONSE: "When evaluating a job offer, consider these factors: 1. Compensation: Is the salary competitive? 2. Benefits: Health insurance, 401k, PTO 3. Culture fit: Does the company align with your values? 4. Growth potential: Room for advancement? 5. Work-life balance: Reasonable hours? Make a pros/cons list and trust your gut." Sounds helpful. Misses the questions that actually matter.
With GigaThink
10 PERSPECTIVES YOU MIGHT MISS: 1. FINANCIAL: Total comp? 401k match? Equity? 2. CAREER: Where do people go after 2-3 years? 3. MANAGER: Your manager = 80% of job satisfaction 4. TEAM: Why is this role open? Growth or exit? 5. COMPANY: Growing, stable, or declining? 6. WORK-LIFE: Glassdoor hours vs stated hours 7. LEARNING: Will you grow faster here? 8. LEVERAGE: Do they NEED you or just WANT you? 9. OPPORTUNITY: Best option or first option? 10. GUT CHECK: Relief or dread when accepting?
What GigaThink catches: The angles you forget when you're excited about an opportunity.

LaserLogic

Spot Bad Reasoning
The blind spot: Arguments that sound wise but hide flawed logic.
Without LaserLogic
TYPICAL AI RESPONSE: "The conventional wisdom is that buying is better than renting long-term because: • You build equity instead of 'throwing money away' • Real estate historically appreciates 3-5% annually • Mortgage payments are fixed while rent increases • Tax benefits from mortgage interest deduction • Pride of ownership and stability If you can afford a down payment and plan to stay 5+ years, buying is usually the smarter financial choice." Sounds reasonable. Hides 4 major logical flaws.
With LaserLogic
HIDDEN ASSUMPTIONS EXPOSED: 1. FALSE EQUIVALENCE Rent = 100% goes to housing Mortgage = 60-80% goes to INTEREST (not equity) 2. MISSING VARIABLES - Down payment could earn 10%/yr in S&P 500 - Transaction costs: 6% realtor fees - Maintenance: 1-2% of home value annually 3. ASSUMES APPRECIATION "Houses always go up" — ask 2007 buyers 4. IGNORES FLEXIBILITY Rent: 30 days to leave Own: 6+ months to sell VERDICT: "Renting is throwing money away" is OVERSIMPLIFIED. Breakeven = 5-7 years minimum.
What LaserLogic catches: Cliches that sound wise but hide bad math.

BedRock

Find the Simple Answer
The blind spot: Overwhelmed with options, missing what actually matters.
Without BedRock
TYPICAL AI RESPONSE: "For optimal health, consider a holistic approach: • Nutrition: Balanced diet with whole foods, consider Mediterranean or keto • Exercise: 150 min moderate or 75 min vigorous weekly • Sleep: 7-9 hours in a dark, cool room • Stress: Meditation, journaling, breathing exercises • Supplements: Vitamin D, omega-3s, magnesium • Biohacking: Cold showers, red light therapy • Fasting: Try 16:8 intermittent fasting • Hydration: 8 glasses of water daily Start with small changes and build habits gradually." 12 things to optimize. But what actually moves the needle?
With BedRock
FIRST PRINCIPLES (Research Consensus): What actually moves the needle? 1. Sleep: 7-9 hours (most ignored, highest impact) 2. Movement: 150 min/week moderate OR 75 min vigorous 3. Nutrition: Mostly plants, enough protein, not too much THE 80/20 ANSWER: If you do ONLY these three things: 1. Sleep 7+ hours (non-negotiable) 2. Walk 30 min daily 3. Eat one vegetable with every meal → You'll be healthier than 80% of people. THE UNCOMFORTABLE TRUTH: You probably already know what to do. The problem isn't information, it's execution.
What BedRock catches: The simple answer hiding under complicated advice.

ProofGuard

Verify Before You Believe
The blind spot: Acting on "facts" you never verified.
Without ProofGuard
TYPICAL AI RESPONSE: "Yes, staying properly hydrated is crucial for health: • Aim for 8 glasses (64 oz) of water daily • Hydration improves energy, skin, and cognition • Dehydration causes headaches and fatigue • Drink more if exercising or in hot weather • Watch for signs: dark urine means drink more The '8x8 rule' is a good baseline for most adults. Keep a water bottle with you as a reminder to stay hydrated throughout the day." Confident advice. But where does "8 glasses" actually come from?
With ProofGuard
CLAIM: "Drink 8 glasses of water a day" SOURCE 1: British Medical Journal (2007) "No scientific evidence for 8x8 recommendation" → Origin traced to 1945 misinterpretation SOURCE 2: Mayo Clinic (2022) → Adequate intake varies by individual → TOTAL fluids (includes food), not just water SOURCE 3: National Academy of Sciences "Most people meet hydration needs through thirst" → No evidence of widespread dehydration VERDICT: MOSTLY MYTH • "8 glasses" has no scientific basis • Food provides 20-30% of water intake • Coffee/tea count toward hydration • Your body's hydration sensor: thirst PRACTICAL TRUTH: Drink when thirsty. Check urine color.
What ProofGuard catches: Widely-believed "facts" that aren't actually true.

BrutalHonesty

See Your Blind Spots
The blind spot: You love your plan and can't see its flaws.
Without BrutalHonesty
TYPICAL AI RESPONSE: "Starting a YouTube channel can be rewarding! Here's how to succeed: • Find your niche: What are you passionate about? • Be consistent: Post 2-3x per week minimum • Invest in quality: Good audio > good video • Engage with your audience: Reply to comments • Use SEO: Optimize titles, tags, descriptions • Be patient: Most channels take 1-2 years to grow • Collaborate: Partner with similar creators Many successful YouTubers started as a hobby and grew it into a full-time income. The key is persistence and genuine passion for your content." Encouraging! But what are the actual odds of success?
With BrutalHonesty
UNCOMFORTABLE TRUTHS: 1. THE MATH DOESN'T WORK FOR 99% • Median YouTuber income: $0 • Top 3% of channels get 90% of views 2. TIME INVESTMENT IS MASSIVE • 1 quality video = 10-40 hours • That's a part-time job for months with no pay 3. YOU'RE COMPETING WITH PROFESSIONALS • MrBeast has a 100-person team • Your "side hustle" vs their full-time career 4. BURNOUT IS THE ACTUAL OUTCOME • 95% of channels quit within first year HONEST QUESTIONS: • Can you do 20+ hrs/week for 2 years with zero return? • Why YouTube? (Newsletter/podcast may be easier) • Is this for money or creative expression? IF YOU STILL WANT TO DO IT: • Make 10 videos before "launching" • Treat it as hobby, not business, until proven
What BrutalHonesty catches: The gap between your optimistic plan and reality.
How It Works

The 5-Step Process

Every deep analysis follows this pattern.

1. DIVERGE (GigaThink)

Explore all angles before narrowing down

→

2. CONVERGE (LaserLogic)

Check logic, detect fallacies, find flaws

→

3. GROUND (BedRock)

First principles, simplify to what matters

→

4. VERIFY (ProofGuard)

Check facts against sources, triangulate claims

→

5. CUT (BrutalHonesty)

Be honest about weaknesses and risks

✓
Divergent → Convergent

Explore widely first, then focus ruthlessly

Abstract → Concrete

From ideas to first principles to evidence

Constructive → Destructive

Build up possibilities, then attack your own work

Profiles

How Thorough Do You Need?

Choose your depth based on the decision's importance.

--quick
~30 sec
Daily decisions
"Should I buy this?"
--balanced
~2 min
Important choices
"Should I take this job?"
--deep
~5 min
Major decisions
"Should I move cities?"
--paranoid
~10 min
High-stakes
"Should I invest $100K?"
What Developers Say

Built for Skeptics

Engineers who've integrated ReasonKit into their workflows

"I was skeptical another reasoning framework would add value. Then I ran my first benchmark—literally 50x faster than my LangChain setup. The Rust core isn't marketing fluff."

Marcus Kim
Marcus Kim
ML Engineer @ Synthesia
@marcuskim_ml

"The BrutalHonesty tool caught an edge case in our recommendation engine that 3 senior engineers missed in code review. Now it's part of our CI pipeline."

Sarah Rodriguez
Sarah Rodriguez
Tech Lead @ Shopify
github.com/srodriguez

"We replaced 2,000 lines of custom prompt engineering with 50 lines of ReasonKit config. Same accuracy, 10x less maintenance. Should've switched months ago."

James Chen
James Chen
Principal Engineer @ Stripe
@jchen_code
Pricing

What would 18x better reasoning be worth?

Start free. Upgrade when it pays for itself.

Core
Open source forever
$0 forever
  • All 5 ThinkTools
  • PowerCombo (full pipeline)
  • Local execution
  • CLI interface
  • Apache 2.0 licensed
  • Community support
Install Free
Most Popular
Pro
For serious decision-makers
$5 /week
billed weekly
  • Everything in Core
  • Cloud API access
  • Advanced reasoning modules
  • Team dashboard
  • Priority support
  • Usage analytics
Start Free Trial
No credit card required
Enterprise
For teams at scale
  • Everything in Pro
  • Unlimited usage
  • SSO/SAML
  • On-premise deployment
From the Blog

Latest Insights

Featured

Why AI Needs Structured Reasoning: The Case for Protocols Over Prompts

AI gives you answers fast. But how do you know they're good? Most LLM responses sound confident but skip the hard questions. We built ReasonKit to fix that: five tools that force AI to think systematically, explore all angles, and expose its assumptions.

December 23, 2025 8 min read Read Article

Think Deeper. Reason Better.

Try It Free