A Witch Hat Atelier spell simulator with a real parser behind the magic
A fan-made browser-based Witch Hat Atelier spell simulator.
At a glance
- What is it?
- This fan-made browser project turns freehand spell diagrams into parser diagnostics, an AST, an intermediate representation and animated canvas effects, using the manga's grammar of sigils and signs. It is an experimental prototype, honest about its limits, with community Discord and docs for extending the dictionary.
- Who is it for?
- The spell simulator fits Witch Hat Atelier readers who want the manga's diagram magic to actually run, and developers curious about freehand diagram recognition compiled into behaviour, who get a tested, documented, contract-publishing prototype to play with or fork.
- Can I use it commercially?
- Yes. MIT is a permissive licence: you can use, modify and sell software built on it, as long as you keep its copyright and licence notices.
- Is it still maintained?
- Yes. The repository last received commits 112 days ago.
- What is it written in?
- Mainly JavaScript, according to GitHub's language statistics.
Answers come from the project's GitHub data, last synced on September 17, 2026, and from our analysis. They are not legal advice.
DEEP OPEN-SOURCE ANALYSIS
Drawing spells, compiled for real
Witch Hat Atelier is a manga whose magic system is unusually procedural: spells are diagrams, drawn with a ring, sigils for elements, and signs that modify the effect. This project takes that premise literally and builds a browser simulator where a freehand drawing becomes parser output, compiled spell behaviour, and animated canvas effects.
The pipeline is the interesting part, because it is a real one. You draw a diagram on a paper-like canvas; the app detects the enclosing ring and distinguishes prepared from active spells; it recognizes dictionary-backed primary sigils for fire, water, wind, earth and light; it recognizes signs that modify direction, levitation, convergence, force, spread, focus, range, duration and stability; and it produces parser diagnostics plus two inspectable artifacts, a glyph abstract syntax tree and a spell intermediate representation, before rendering the animated element effects.
That is the structure of a compiler front end wearing a wizard hat, and it is what separates this from a drawing toy with sound effects. The failure modes are diagnostics, not silence, and the outputs are structured enough to build on.
Honest limits, stated up front
The limitations section is long, specific and unusual for a fan project. One enclosing ring at a time, with multiple rings detected as unsupported. One primary sigil per spell, with multiples likewise rejected. Recognition is based on local stroke templates, so it works best with clean, deliberate drawings, and the recognizer is imperfect: some valid-looking drawings fail to match, and rough ones need redrawing.
The dictionaries cover a small fan-made subset of sigils, signs and observed spell ideas, the visual effects are interpretive canvas animations rather than reproductions of the manga's artwork, and raster images can serve as visual references but true stroke order cannot be recovered from them. Closed but invalid diagrams show diagnostics rather than falling back to some other element.
Every one of those sentences is a design boundary rather than an apology, and together they tell you the author knows exactly what was built. The project status statement completes the picture: an experimental prototype, with issues and pull requests reviewed when time allows, no commitment to long-term maintenance, and an explicit invitation to fork.
Run it, test it, extend it
Running locally is the standard pair:
npm installnpm startThen open:
http://127.0.0.1:5173/The test suite runs headless:
npm testReference tools ship inside the app for the part that matters most in template recognition, building and inspecting the templates themselves:
/tools/strokeTemplateMaker.html
/tools/strokeTemplateViewer.html
/tools/sigilSignDetectorLab.html
/tools/spellEffectLab.htmlA stroke template maker and viewer, a detector laboratory for testing recognition, and an effect lab for tuning visuals, which is the toolchain a project like this needs if anyone else is going to contribute sigils.
The documentation set covers the same ground in prose: dictionary authoring, the parser and spell semantics rules, the parsed glyph output contract, the compiled spell output contract, and notes on the effect renderer. Publishing output contracts for a fan simulator is the kind of rigour that makes forks productive, and the community Discord exists for people taking it in different directions.
The fan-project posture, done correctly
The legal posture is stated carefully and early: an unofficial fan-made project for learning, experimentation and appreciation, not affiliated with, endorsed by or sponsored by the creators, publishers, licensors or production partners. The names, artwork, symbols and trademarks belong to their rights holders, and the sigils, signs, terminology and effects in the project are partial fan references and interactive interpretations, not official assets or canonical rules.
That paragraph is the difference between a fan project that survives and one that receives a letter. Nothing here reproduces protected artwork; the recognition targets are stroke templates, the effects are original animations, and the dictionaries are explicitly fan-made subsets.
The sample spell layouts in the dictionary panel deserve a mention as the onboarding path: rather than documenting the grammar in the abstract, the app shows drawable examples you can copy, which is how a visual language should teach itself.
Who it is for, and the neighbours
The audience is the intersection of two groups: readers of the manga who want to play with its magic system as a system, and programmers interested in a working example of diagram recognition compiled to behaviour, which is a genuinely hard interaction problem dressed in the friendliest possible costume.
The neighbours on the fiction side are the fan wikis documenting the manga's spell grammar, which catalogue but do not execute. On the engineering side, the nearest relatives are handwriting and gesture recognizers, which solve the matching problem without a domain grammar, and diagram editors, which have the grammar without freehand input. This project needs both at once, ring detection, dictionary lookup, sign parsing, then compilation, and its limitation list is largely the story of how hard that intersection is.
As a review judgement: for a prototype with no release tags, pushed last on 2026-05-31, it is unusually complete, tested, documented and honest, and the fork invitation is the right governance for fan work. Whether it grows depends on the community that formed around it, which is exactly what the Discord and the contracts are for.
Editorial conclusion
The spell simulator fits Witch Hat Atelier readers who want the manga's diagram magic to actually run, and developers curious about freehand diagram recognition compiled into behaviour, who get a tested, documented, contract-publishing prototype to play with or fork. It does not fit anyone needing a maintained product, since the author explicitly commits to nothing long term, or anyone expecting canonical spell rules and official artwork, which the fan notice correctly disclaims. Verify first: that your drawings match the template recognizer's tolerance for deliberate strokes, the dictionary's coverage against the sigils you care about, and the output contracts if you plan to extend rather than just play. The licence is MIT, and the last push was on 2026-05-31.
Frequently asked questions
What does the Witch Hat Atelier spell simulator do?
It turns freehand spell diagrams into parser diagnostics, a glyph AST, a compiled spell representation and animated canvas effects, recognizing dictionary-backed sigils for five elements and modifier signs for properties like direction, force, range and duration.
Is the spell simulator an official Witch Hat Atelier product?
No. The project states it is unofficial and unaffiliated, that the manga's names, artwork and trademarks belong to their rights holders, and that its sigils, terminology and effects are fan references and interactive interpretations, not official assets or canonical rules.
Can the simulator recover a spell from an image?
No. Raster images can be used as visual references, but the app cannot recover true stroke order from an image, and recognition is based on local stroke templates that work best with clean, deliberate freehand drawing.
Community notes