I built a free tool to analyze your chess openings repertoire weaknesses

The Real Problem with Opening Study

Most chess players have a vague idea of their repertoire and a very precise idea of what they wish their repertoire was. You "play the London". You "answer e4 with the Caro". Then you pull up your last 200 games and discover that 40% of them are some Indian setup you stumbled into, you've faced the Sicilian 18 times and scored 22%, and your beloved Caro shows up in 9 games out of 200.

Studying openings without that picture is guesswork. You spend an evening on a sideline you'll see twice a year, while the line that actually decides half your games quietly bleeds rating points.
The fix isn't a course. It's data: what do you really play, against what, and how does it score?

How OpeningScanner Works

OpeningScanner streams your games from Lichess or Chess.com, classifies every one against the full ECO database, and gives you a dashboard of your real repertoire. Four steps:

Step 1: Drop in a username
Lichess or Chess.com handle, color filter, time control, max games. Scans up to 2000 games, newest-first. No login, no account, nothing to install.

Step 2: Stream and classify
Games stream in live, NDJSON from Lichess or monthly archives from Chess.com. Each game is replayed up to 24 plies and matched against the Lichess chess-openings database (about 3000 entries, EPD-keyed so move-order transpositions land on the same opening). You watch the count climb in real time and you can cancel mid-stream.

Step 3: Drill into the dashboard
Openings grouped by family, sortable by frequency or score. Click in and you get the move tree past the ECO line: every continuation you've actually played, with W/D/L and average opponent rating at every node. An interactive board on the side, arrow keys to step through.

Step 4: Find the gaps and the weak spots
Two reports built on the same data. Weak spots: openings where you score worst, weighted by how often you play them (the ones actually costing you points). Gaps: popular openings you barely touch, scoped to whatever family you're drilling into. Export filtered PGN of the raw games or a study-ready PGN with every continuation as a nested variation, ready to drop into a Lichess study.

What Actually Surprised Me

I expected my repertoire to be narrower than I thought. It was the opposite. I had 60+ distinct openings show up across 1500 games, most with single-digit counts, and the long tail was where almost all the bad scores were hiding. The "main lines" I was studying were fine. The damage came from the 30 lines I'd never decided how to play and had been winging for two years.

Try It

OpeningScanner is at https://openingscanner.com. Free, open source (MIT, code on GitHub), no account, nothing stored on a server: games stream straight from Lichess and Chess.com to your browser, classification runs in a Web Worker, the last scan is cached in IndexedDB so a refresh doesn't lose it. Available in 13 languages.

If you find a broken handle or a weird classification, the repo takes issues and PRs.