Should I vibe code
Native Mac interface design with prototypes, libraries, and collaboration
Vector editors are a genuinely hard software category and always have been.
?
Their verdict, the Standard price and the build-time estimate come from their entry, MIT-licensed. Checked 2026-08-03.
?
Our verdict, the regret score and everything below it. Editorial and unsponsored — nobody can pay to be moved.
The honest answer
why the verdict is what it is
You can build a shape tool. Getting to boolean operations, prototypes and shared libraries is the part where you discover why this category has three survivors.
What actually breaks
not "if". the specific failures.
- Boolean path operations, which are computational geometry with degenerate cases that crash or produce nonsense
- The vector editing interactions — bezier handles, snapping, alignment guides — which are judged against muscle memory built over years
- Symbols and component overrides, a nested inheritance model that is the reason anyone uses a design tool at scale
- Undo across a canvas with grouped, nested, multi-select operations
- Rendering fidelity, so blend modes, blurs and shadows match what the browser or the exporter will actually produce
The canvas works, shapes draw, and you are pleased until you implement subtract on two overlapping paths that share a tangent. The result has a stray segment and a winding direction that fills the wrong region. You read about the even-odd rule, then about numerical tolerance for near-coincident points, and you are now debugging floating point in computational geometry — a field with a literature — several weeks into a project you started because you wanted nicer alignment guides.
Is that you?
the verdict is a default, not a law
- It is built on an existing vector engine rather than implementing geometry yourself
- It is a viewer or a narrow generator, not an editor
- You are doing it to learn graphics and shipping is not the point
- You are implementing boolean path operations from scratch
- The goal is to replace a tool people use professionally every day
- Designers other than you must adopt it
- There is no import or export in a format other tools read
If you build it anyway
the checklist, then the prompt that enforces it
- Build on an existing vector and geometry library. Boolean operations and path offsetting have a literature and a long tail of degenerate cases.
- Decide early whether this is an editor or a generator. Generators are tractable; editors compete with muscle memory.
- Implement undo as a command history from the first commit — retrofitting it onto a canvas is a rewrite.
- Support import and export of an interchange format early, so the work is never trapped and adoption is not all-or-nothing.
- Test rendering against the target — browser, print, or exporter — rather than against your own canvas.
- Symbols and overrides are a large subsystem. Either commit to them deliberately or say they are out of scope.
Before you build a vector design tool, apply these and push back if I ask you to break them. 1. Ask me whether this is an editor or a generator. If it is an editor, tell me I am competing with software people use daily and have muscle memory for, and that the bar is interaction quality rather than feature count. 2. Do not implement boolean path operations, path offsetting or stroke outlining yourself. Use an established geometry library and tell me these have well-known degenerate cases involving coincident points and numerical tolerance. 3. Implement undo as a command history from the first commit. Every canvas operation, including multi-select and nested group edits, must be a reversible command. 4. Support importing and exporting SVG early, so work made in the tool is never trapped and can round-trip with other software. 5. Test rendering fidelity against the actual target — browser, PDF, or exporter — rather than assuming the canvas is authoritative. Blend modes, blurs and shadows differ between renderers. 6. If symbols or components are in scope, design the override and inheritance model before building them, and tell me it is a subsystem rather than a feature. 7. Get snapping, alignment guides and keyboard nudging right early. They are what makes a design tool feel usable within the first minute. 8. Out of scope unless I ask: prototyping, collaboration, plugins, developer handoff, shared libraries. 9. Give me an honest estimate before I begin.
That one keeps you out of trouble. For the prompt that actually builds it, canivibecodeit.com has one.
their build prompt ↗Or don’t build it
the boring option, and the way back out
Almost certainly. $12 a month against a mature vector editor is not a close comparison, and the parts that look simple — snapping, booleans, symbols — are the parts with the deepest tails. Building a narrow generator that outputs SVG for something you do repeatedly is the version worth doing.
$12/mo is cheaper than your weekend.
Use SVG as the native or interchange format so files are readable by other tools from the start. In a category where abandonment is the likely outcome, the work surviving the tool matters more than any feature you might add.
Open-source collaborative design and prototyping platform.
Self-hosted photo organization and search.
Questions
Why are boolean operations so much harder than they look?
Because the easy cases — two clearly overlapping rectangles — are trivial and the real cases are not. Paths that touch at a single point, share a tangent, self-intersect, or have segments closer together than floating point can distinguish all produce results that are wrong rather than failed. Robust implementations handle this with exact arithmetic or careful tolerance strategies, and both are specialist work.
What's the generator version of this project?
Something that produces SVG from data or parameters — chart artwork, social images, repeated diagrams — with no interactive editing at all. It skips undo, snapping, booleans and symbols entirely, and for many people it is what they actually wanted: not a design tool, but a way to stop making the same graphic by hand.
Every week, someone ships something they shouldn’t have.
New verdicts, the worst thing that landed in the trap, and the occasional incident report. No other email, ever.
last reviewed 2026-08-03 · verdict is editorial and unsponsored · shared entry data from canivibecodeit under MIT · not legal advice