How to get chess puzzles from your own games
You can get chess puzzles from your own games in five ways: Lichess's Learn from your mistakes and Chess.com's Retry turn the mistakes of one analyzed game into find-the-move exercises; you can save positions by hand for free; and some tools, Chess Coach's own-mistake drills among them, collect the errors from many of your games into one set. As of October 26, 2026, the per-game options are free on Lichess and limited to one full review a day on a free Chess.com account.
The options compared
| Option | What it uses | Scope | Cost (October 2026) | Source |
|---|---|---|---|---|
| Lichess "Learn from your mistakes" | Your analyzed game | One game at a time | Free, unlimited; server analysis 40 games/day | Lichess features |
| Chess.com Game Review "Retry" | Key moves in your reviewed game | One game at a time | 1 full review/day free; unlimited on paid plans | How does Game Review work? (updated Sep 25, 2026) |
| Manual collection | Positions you save yourself | Any number of games | Free | — |
| Third-party sites (e.g. Chessiro) | Imported Chess.com/Lichess games | Many games | Varies | Chessiro blog (Jul 19, 2026) |
| Chess Coach | Your finished Chess.com/Lichess games | One game (Practice mistakes) or 3–30 games, recent or from a period (drills) | Practice: Free plan; drills: 14-day free trial, Plus/Pro | This page |
One common confusion: the big puzzle trainers are not personal. Chess.com explains that its engine finds blunders in games played on the site (and other sources) to create puzzles, so "any position from your own games could become a puzzle" in the shared pool (Chess.com Help Center, September 26, 2024). Lichess lists "tactical puzzles from user games" too. Both are excellent practice, but they are other people's mistakes.
The per-game modes on Lichess and Chess.com
Lichess. Open your game, request a computer analysis, and a "Learn from your mistakes" button appears. Lichess introduced it in December 2016 as a way to train from your own games: it takes you to each mistake and asks for a better move, and only shows the answer if you ask. It chooses the mistakes by how much your winning chances dropped, so it skips small slips in positions that were already decided (Lichess blog, December 22, 2016).
Chess.com. Inside Game Review, key moves that could have been better come with a Retry option, which replays the position so you can try to find the best move (Chess.com Help Center, updated September 25, 2026).
Both work very well right after a game. Their limit is that each session covers one game, so a pattern that costs you a piece every few games is hard to spot.
A real example: the same capture, missed twice
This one is from my own account (nacsol3), May 14, 2024: a 20-minute game with White against NeverGonnaCheckYouUp (1030), who played the Englund Gambit. My bishop on f4 and their bishop on d6 were looking at each other, and the open d-file meant my queen also attacked their bishop. Whoever traded first would come out ahead.
Stockfish rates this position +7.15 for White (completely winning) with the queen taking on d6: after the queens come off, I keep an extra bishop. I played the bishop to b5 instead, pinning their knight (a pin is when a piece can't move without exposing something more valuable behind it). It felt clever, but it ignored the capture on the other side: Black took my bishop on f4, the queens traded on d8, and now I was the one a piece down. The engine score after my move was -5.26, and I resigned ten moves later.
The part that makes it a perfect puzzle: one move earlier, both of us had the same capture available, and both of us missed it. The engine flagged my previous move and my opponent's reply as blunders for that reason. A position like this is worth solving again next week, not just reading once.
How to build your own set for free
- After each analyzed game, copy the position before your biggest mistake (the FEN, a one-line text code for the position) into a Lichess study or a notes file.
- Write the question, not the answer: "White to move. What did I miss?"
- Solve the set again a few days later without the engine. Keep the ones you miss; drop the ones you find instantly.
It's slow, but free and it works on any site. Our guide on how to analyze chess games as a beginner shows how to pick the right positions.
How Chess Coach does it
Chess Coach is a connector for Claude and ChatGPT that reviews your finished Chess.com and Lichess games with Stockfish.
- Practice mistakes (all plans, including Free) is a mode on the review board: the positions where you went wrong come back so you can find the better move yourself.
- Own-mistake drills (14-day free trial, Plus and Pro, the
training_drillstool) collect the mistakes and blunders from up to 30 games, your most recent ones or those of a period such as last month's rapid games, with the blunders first. Your assistant shows them one at a time on a board without the answer, asks for your move, then shows the solution with arrows and a plain-word explanation. The ones you miss come back at the end.
A prompt to try: "Make drills from my last 10 Chess.com games." Every account gets drills free for its first 14 days; after that Plus costs $3.99 a month. See the AI chess coach page for what each plan includes, or connect Chess Coach to try the free review first.
Frequently asked questions
Can Lichess make puzzles from my own games?
Inside one game, yes: after a computer analysis, the "Learn from your mistakes" button walks you through your errors and lets you find the better move. Lichess's features page lists it as unlimited on free accounts. Its puzzle trainer itself uses positions from games across the site, not a personal set.
Does Chess.com turn my games into puzzles?
Game Review has a Retry button for key moves that could have been better, so you can replay those positions. Chess.com also says any position from members' games could end up in its general puzzle pool, but that pool isn't built around your mistakes.
Are puzzles from your own games better than random puzzles?
They cover a different need. Random puzzles show you many patterns; your own positions show the exact ones you miss in real games. A mix of both works well.
How many own-mistake puzzles should I do?
A handful at a time is enough. What matters is coming back to the ones you got wrong a few days later, until you find the move straight away.