Find the first line where an algebra solution goes wrong.
Type a problem and your working, one step per line. Each step is checked against the one before it, the first mistake is named, and you get hints rather than the answer. Worked solutions, graphs and practice sets are included.
The app itself, running inside this page. Anything you do here is saved in your browser.
Ask it to do things
Press Ctrl+K in the app (Cmd+K on a Mac) and describe what you want. The model chooses from the app's own functions, runs them in order and tells you what changed. Some examples:
| Check my work on this equation | Verifies each line and finds the first error. |
| Show me a worked solution to 2x^2 - 3x + 1 = 0 | Solves it step by step, with each step verified. |
| Give me five problems on my weakest mistake | Starts a practice set aimed at that misconception. |
Instructions you save in the same box, such as "keep it short" or "I'm a beginner", are added to every AI request the app makes.
Features
- Line-by-line verification of algebra
- Expressions, equations and 'x = 2 or x = 3' answer sets.
- Numerical equivalence engine
- Random-point testing, proportionality test, root finding with bisection + ternary search.
- Local error classification
- Sign flip, constant offset, scale factor, lost/extra solutions.
- KaTeX-rendered steps
- With / status and evidence.
- AI misconception tagging against a
- Fixed taxonomy (exponent over addition, partial division, sign moves …).
- Progressive hints
- 3 Socratic nudges, then the correct step.
- Persistent misconception profile that grows
- With errors and decays with clean solutions; targeted practice generator.
- Solver page
- Worked solutions for linear and quadratic equations (factoring when roots are rational, the quadratic formula otherwise) and polynomial expansion, with every step run back through the checker.
- Graph page
- Plots both sides of an equation or an expression on a canvas, marks the solutions, and auto-fits the window to the roots.
- Practice page
- Generated problem sets per topic, or aimed at your weakest misconception; each answer must be a finished form and every step must verify.
Runs in your browser
Parsing, verification, root finding, error classification, rendering, profile. None of this needs a network connection or a model.
Uses a language model for
Misconception diagnosis grounded in verifier evidence (JSON, fixed taxonomy). Practice-problem generation aimed at the student's weakest misconception.
Notes
- Cost
- Free. Each account has a daily allowance of AI requests.
- Your data
- What you make stays in this browser on this device. Your account is only used for signing in and for the built-in AI. You can export a backup or clear everything from Settings.
- AI access
- Signed-in users get the built-in AI with nothing to set up. You can also use your own Groq or OpenRouter key under Settings.
- Source
- The code and tests are on GitHub under the MIT license. The tests also run in your browser.