Not an AI pretending it found the perfect play. Three real game questions, each with a source, a useful answer, a limitation, and a question for the community.
Evidence matters. “Documented” means we have a published rule or source. “Player reported” means someone described what happened, not that GamePilot reproduced it. “Untested” means there is no independent live-game verification. No real-time gameplay AI has been validated here.
It counts draw-deck cards left, not cards you started with.
Card rule
01 — Slay the Spire 2 / 杀戮尖塔2
0 HP lost ≠ 0 resources lost
The player report: In CombatSolver #239, the player said a solver chose EndTurn despite having three attacks that, according to their later comment, could kill the Fat Gremlin. The encounter would end either way, but escape changes its gold reward.
Documented rule: The published STS2 v0.107.1 change log says that if the Gremlin Merc hadn't stolen any gold and Fat Gremlin escaped, the player now receives 50% of normal encounter gold rather than none. It does not mean that every escaping enemy halves rewards.
Why it matters: A line valued only by HP loss can wrongly treat kill and escape as equivalent. Gold recovered, escape versus kill, energy, and health should be considered separately. Example arithmetic: if a hypothetical normal reward were 40 gold, the documented escape condition awards 20. Forty gold is an illustration, not this player's observed reward.
Check each alleged lethal attack's energy cost, target and effective damage.
Compare kill and escape outcomes: HP lost, actual reward, stolen gold recovered, and execution legality.
Reproduce any proposed alternative in the right environment before calling it a verified fix.
Limit: GamePilot did not run this game or reproduce the solver defect. The three-attack lethal claim is from the player. Our public diagnostic reply suggests tests, not a confirmed fix.
Your call: Should a solver prefer preserving HP or earning more gold when both lines end the encounter?
Player's confusion: An official-forum report argues that Entangle's generalized “Slow” wording suggests attack and cast speed should change, yet only enemy movement was observed to slow.
Reference answer: The community mechanics page describes the current documented Entangle effect as movement-speed slow only. Data listings describe a 15% slow per attached vine, but do not by themselves establish attack/cast slowdown or the precise way multiple vines stack.
Practical decision: Treat this as movement control, not guaranteed reduction in attacks per second. If your build depends on avoiding hits, test rather than counting a generic tooltip as an attack-speed debuff.
A useful four-step reproduction
Note the exact client patch, Entangle gem level, supports and monster type.
Observe several attacks with no other slows active; record their timings.
Apply vines, then separately measure movement and attack/cast intervals.
Repeat and share anonymized timestamps, not a claim based on one animation.
Limit: The original report is from December 2025. GamePilot has not repeated this on the October 2026 build; patches may change mechanics.
Your call: Are you seeing movement-only slowdown in the current patch? Which build, monster and supports?
Compare ordinary-card plays, then enter the next state.
The Hand Coach compares every supported selection from up to eight manually entered cards and explains the immediate score. Only ordinary cards without Jokers or modifiers are supported. Screenshot AI is unavailable; no paid checkout is active.
Balatro scoring explained / Synthetic worked example / 17 October 2026
You played five cards. Three scored zero Chips.
AS AH KD QC 2S → Pair → 64 points. With level-1 hand values, ordinary unmodified cards and no Jokers or special effects, only the two Aces score: (10 base Chips + 11 + 11) × 2 Mult = 64. The King, Queen and 2 do not add their card Chips to this Pair. Playing just AS AH also scores 64 under these assumptions.
This is a scoring lesson, not advice to spend or keep those extra cards. Playing unscored cards can change your next draw; this tool does not evaluate that future value. It compares immediate scores only.
Enter AS AH KD QC 2S, target 60, leave level-1 base values, and confirm the supported rules. This is an authored synthetic demonstration, not a player's run or testimonial. AI-assisted research and writing; unofficial GamePilot project. No screenshot recognition or paid checkout.
If you try your own supported position, optionally share the entered cards, predicted score, actual in-game score and one confusing point through the existing free review link in the tool. No review or positive feedback is required.
03 — Balatro / 小丑牌
Blue Joker is not a fixed +104 Chips.
Documented ability: Blue Joker adds 2 Chips for every card remaining in the draw deck when it scores. It does not count your hand, played cards or discarded cards. The bonus falls as the deck is drawn down, so early-Blind value differs from late-Blind value.
Rule:Blue Joker Chips = 2 × cards still in draw deck. This is an additive Chips effect; multiplying that by Mult is only a simplified score estimate when other interactions are excluded.
Try the numbers yourself
Blue Joker contributionApprox. +840 score at 12 Mult, ignoring other interactions.
The calculator runs locally in your browser and sends no input anywhere. The initial numbers are illustrative, not observations from a player's run.
Manual input only. Screenshot review, if requested, is human-reviewed and AI-assisted during active sessions. No automatic recognition or continuous tracking has been validated. Calculator and research reviews are free; no paid checkout is active.
Decision checklist
Count what remains in your draw deck now, not the whole deck.
Compare its +Chips contribution to a replacement Joker's actual effect at your current Mult; a +Mult or ×Mult Joker is not directly comparable by printed value alone.
Recalculate after more drawing or discarding.
Include synergy, card modifications, boss effects and the rest of the build before selling.
Limit: The published ability supports the formula. Without a player's hand, Jokers, upgrades and Blind, there is no defensible universal “sell it” recommendation.
Your call: How many cards were actually left in your deck when Blue Joker stopped helping?
Bring one hard decision. Let's compare the evidence.
Send the game and patch, a public position link or redacted text description, and one sentence about the choice. No Steam password, no game installation, no payment.
Important: the current anonymous Google Form does not upload files directly. You can paste a previously public game-state/screenshot URL or write a text summary. If you'd like a response, optionally leave a public thread link or contact channel; anonymous entries without one cannot receive private replies.
GamePilot is an experimental human-reviewed, AI-assisted research desk, not a validated live autopilot. Response times are not guaranteed. We won't republish your screenshots as case studies without asking. You are welcome to challenge or correct any answer.
How we mark confidence
Label
Meaning
Documented
Mechanic or patch documented, with conditions and a source.
Player-reported
An independent player described or demonstrated an outcome; not yet our reproduction.
Derived
Math or inference from stated rules, with assumptions.
Untested
Hypothetical action or fix requiring a real check.