Features
Everything MindFrame does — and why each part matters
MindFrame is not a quiz app with a score. Every feature is designed around a specific gap in how people think — and a precise mechanism for closing it.
Modes & games
19 training modes and 6 games.
Each mode targets a specific metacognitive skill, and each game centres one. Every mode and game is free. A Season Pass adds deeper AI coaching, certifications and a Season Report.
Calibration Lab
FreeRate confidence on every answer. Track the gap between stated certainty and actual accuracy. The top forecasters in the world train precisely this — and it's measurable inside 50 attempts.
Prediction Arena
FreeLog quantitative predictions on real-world events with timestamps. Your Brier score accumulates; your calibration curve becomes a public receipt of your judgment over time.
Uncertainty Engine
FreeA high-ambiguity mode where the conclusion matters less than the proportionality of your reasoning. This is the purest test of judgment under uncertainty.
Bias Hunter
FreeLearn to name the specific bias operating in any argument, recommendation, or decision. Fluency reduces susceptibility by roughly 30% in follow-up tests.
Assumption Audit
FreeTrain the habit of asking what must be true for a conclusion to follow. Assumption Audit targets the invisible premises that make arguments feel stronger than they are.
Thinking Trap Escape
FreeA more intervention-heavy bias mode. Thinking Trap Escape is about actively exiting a cognitive failure pattern, not merely naming it.
Emotional Radar
FreeTrain emotional self-awareness as a metacognitive skill. Emotional Radar focuses on how feelings distort judgment, not on labelling feelings for their own sake.
Attention Ladder
FreeProgressive attention intervals train focus like overload training: start short, hold cleanly, then climb as your window expands.
Focus Guardian
FreeA visual inhibitory-control game that trains sustained attention by making distraction resistance visible, measurable, and progressively harder.
Distraction Gauntlet
FreePractice reading and reasoning while notification-style distractors compete for attention; every interruption creates a measurable cost.
Signal vs Noise
FreeTrain selective attention for evidence quality. Signal vs Noise teaches you to weight the right variables instead of the vivid or distracting ones.
Argument Gym
FreePractice dialectical reasoning by taking a position, facing the strongest opposing case, and scoring how honestly you update or defend.
Contradiction Challenge
FreeTrain the ability to spot incompatible claims, hidden inconsistency, and mutually undermining beliefs before they calcify into bad arguments.
Reframe Forge
FreeTrain the move from distorted interpretation to evidence-based reframe. This mode sharpens emotional regulation by improving the thought structure underneath it.
Strategy Selector
FreeMetacognition is not just noticing errors. It is selecting the best intervention. Strategy Selector trains the mapping from problem pattern to mental move.
Memory of Thought
FreeTrack how your first interpretation shifts or fails to shift as more information arrives. This mode measures anchoring resistance in motion.
Reflection Sprint
FreeShort, structured introspection that trains the habit of naming your own reasoning patterns instead of only reacting to outcomes.
Decision Lab
FreePractice choosing between imperfect options with incomplete information. Decision Lab turns vague judgment into explicit tradeoff thinking.
Metacognitive Scan
FreeMost errors aren't content errors. They're monitoring, control, evaluation, or planning errors. Scan trains you to classify the meta-level failure behind any thinking sample.
ASCENT
Choose a route, scout a clue, and earn your way up the mountain. Save your supplies for the climb that matters.
Trains: Information seeking and answer revisionEvidence Showdown
Spend a limited battery on revealing tests. Rule out failed modules and restore three beacons using only conclusions the evidence supports.
Trains: Choosing informative evidenceRule Shift
Swap neighbouring tiles to restore a relay. When the rule changes, look at the same pieces in a different way.
Trains: Adapting a plan to changing rulesSignal Station
Route active cargo through noisy broadcasts. Choose when to quiet the chatter and repair a misroute before the next packet.
Trains: Selecting relevant informationMemory Heist
Case the vault, hide the plan, and collect its fragments. Return after another room and find out what you remember.
Trains: Retrieval of learned pathsFog Cartographer
Chart a fogged map with a fixed probe budget. Ping wide, dig narrow, and decide when the next look is no longer worth it.
Trains: Deciding where to look and when to stop lookingThe scoring system
3-layer scoring — because right/wrong is the floor
Every session produces three scores. Each one measures something the others miss.
Session result
Bias Hunter — standard session
01 — Outcome
78%
7 of 9 correct
Whether you got it right. The floor, not the ceiling.
02 — Reasoning
64 / 100
Logical structure
Quality of your argument, not just the answer.
03 — Self-awareness
71 / 100
Confidence accuracy
How well your confidence matched your accuracy.
Composite Score
Weighted across all three layers
71 / 100
Example session result. Your scores are tracked and trended across every session.
AI Coaching
Claude analyzes every session
After each session, Claude reads your full session log — every answer, every confidence rating, every reasoning trace — and produces a personalized debrief.
This is not a generic tips list. It identifies the specific pattern behind your errors in this session, names your weakest dimension, and prescribes the exact next mode to train.
- Names your weakest metacognitive dimension from this session
- Surfaces the pattern in your errors — not just the errors themselves
- Prescribes the specific mode to train next
- Tracks whether prior session prescriptions improved your scores
AI Coaching Debrief
Session 7 — Bias Hunter
Cognitive Twin — live model
Predicted blind spot: Overconfidence in domains with strong prior beliefs. Watch Calibration Lab and Contradiction Challenge results.
Cognitive Twin
Your evolving thinking model
The Cognitive Twin is a live model of how you think — built from your session history and updated after every training session. It tracks your five metacognitive dimensions over time, not just today's score.
As the model matures, it begins to predict your blind spots — the specific question types, domains, and emotional conditions where your calibration tends to break down. These predictions sharpen the AI coaching recommendations.
- Tracks all eight Cognitive Fingerprint axes independently across your full session history
- Surfaces persistent patterns, not just recent sessions
- Predicts your blind spots before you encounter them
- Updates after every session — no manual input
Cognitive Fingerprint
A shareable map of how you think
After 20 sessions, your scores across eight cognitive axes stabilise into a recognisable pattern — your Cognitive Fingerprint. It is specific to you, built from your actual performance data, and updated as you improve.
The Fingerprint is a shareable public profile badge — a single-image visualisation of your thinking profile that you can attach to your professional bio, portfolio, or LinkedIn.
- Radar chart across all 8 Fingerprint axes
- Updated automatically as your scores change
- Public profile badge — shareable link and image export
- Side-by-side comparison with anyone you share it with
Cognitive Fingerprint
Built from 24 sessions
72
Calib.
64
Reason
81
Bias
69
Belief
75
Memory
Prediction Journal
Track what you predicted and whether you were right
Log predictions about real-world events with confidence ratings and target resolution dates. MindFrame timestamps each prediction and surfaces it when the resolution date arrives. Accuracy is computed automatically via Brier Score.
- Timestamped entries with confidence level
- Automatic outcome resolution reminders
- Cumulative Brier Score across all logged predictions
- Topic-level accuracy breakdown
Forecasting Arena
Compete on calibration — not just accuracy
The Forecasting Arena presents shared prediction challenges to all members. The leaderboard ranks by Brier Score — so the winner is the most calibrated forecaster, not the most confident one. Compete with others on the dimension that actually matters.
- Shared challenges — same question for all participants
- Ranked by Brier Score, not just % correct
- Weekly and all-time leaderboards
- Topic specialisations (geopolitics, technology, markets)
Adaptive learning
The right mode at the right time
MindFrame's Adaptive Learning Path reads your Cognitive Twin data and surfaces the mode that will produce the highest improvement gain for your current weakest dimension.
You never have to wonder what to work on next. The system knows which dimension is lagging, which mode targets it most directly, and presents it on your dashboard as your recommended next session.
- Reads your Cognitive Twin — not just your last session
- Prioritises by improvement potential, not by what you enjoy most
- Updates the recommendation after every session
- Explains why it chose the mode it did
Recommended next session
Calibration Lab
Free
Your Calibration score has been your weakest dimension for 3 sessions. Calibration Lab specifically targets the confidence-accuracy gap — estimated +8pt improvement at your current trajectory.
Social & competition
Compete on the dimensions that matter
The MindFrame leaderboard ranks by calibration and reasoning quality — not trivia scores. Cognitive Wars are head-to-head matches on specific metacognitive dimensions.
Calibration Leaderboard
Ranked by Brier Score and composite metacognitive score — not just how many questions you answered. The top performers are the most calibrated, not the most aggressive.
- Weekly and all-time leaderboards
- Category-specific rankings
- Percentile benchmarks once the community is large enough to be meaningful
Cognitive Wars
Head-to-head matches on a specific metacognitive dimension. Challenge a colleague or a random opponent — win by outscoring them on calibration and reasoning quality, not speed.
- Async format — no live scheduling required
- Dimension-specific matches
- Team vs. team competitions for workspace accounts
Certifications
Five specialisations — earned, not bought
MindFrame certifications are performance-gated. You cannot pay for them. Each one requires sustained performance across multiple sessions.
Calibration Expert
Brier Score below 0.10 sustained across 25 sessions. Your confidence is a reliable predictor of your performance.
Reasoning Pro
Average AI reasoning score above 85 across 20 sessions. Your arguments are logically sound and evidence-based.
Bias Detector
Identify bias correctly in 90%+ of Bias Hunter challenges across 15 sessions.
Forecasting Specialist
Top 10% Brier Score in Prediction Arena across 50+ tracked predictions.
Metacognition Master
All five session skills above 80 simultaneously for five consecutive sessions.
Team spaces — free, in beta
Team practice spaces are free and in beta: create a team, invite people by code and practise together. There is no paid team plan. There is no paid team plan and no per-seat price, because billing does not support one. Each person uses the free plan or buys their own pass.
Open your team spaceWant a paid team plan?
If you would want a paid team plan, leave your details and we will tell you if one exists.
Join the interest listGet started
Start your first session — free
Free plan includes all 19 training modes, every game, calibration scoring and session analytics. No credit card required.
Answer one challenge. No sign-up.
This is what a real MindFrame attempt looks like — confidence rating included. Your answer becomes one input the AI coach reads.