Should I vibe code
Browser photo editing, templates, branding, and social graphics
Your version exports PNGs. The product is a document you can reopen next year and reshape into thirty formats.
?
Their verdict, the Pro price and the build-time estimate come from their entry, MIT-licensed. Checked 2026-08-04.
?
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
The BeFunky entry already establishes that the operations are library calls and the template library is a licensing department, and none of that changes here. What PicMonkey is really selling sits one level up: it is a filing cabinet with an editor bolted on. Everything you make lands in the Hub as a re-editable design, and the two features people pay Pro for — Brand Kit and Smart Resize — only work because that design is stored as intent rather than as pixels. Smart Resize is the tell. Turning a 1080-square post into a 1500x500 header is not a scale transform; it needs the document to know that this block is a headline with a minimum size, that this shape is a background that may bleed, that this logo has clear space it must keep. Your weekend produces PNGs, which is the one artefact that cannot be resized, rebranded or reopened. Build the editor for fun. Understand that the moment you save your first design in a format only your code understands, you have taken on the least interesting maintenance commitment in software.
What actually breaks
not "if". the specific failures.
- The document format, which you will invent in an afternoon and then owe forever: every design you save is readable by exactly one program, and that program is changing weekly
- Reopening old work after a schema change, which is the moment the Hub stops being an archive and becomes a folder of files that crash the editor
- Smart Resize, because resizing a design is a constraint problem — anchors, minimum type sizes, safe areas, reflow — and a proportional scale produces a header with eleven-pixel body text
- The Brand Kit, which sounds like a colour array and is actually a rule system: which font for headings, which for body, what the logo may sit on, what happens when a palette changes under a hundred existing designs
- Fonts, in both directions — a design that renders correctly for you and in fallback for everyone else, and a typeface licence that does not cover embedding it in exports
- Text layout, the most underestimated component in any design tool: kerning, hyphenation, justification and the label that is one word longer than the template expected
- Upload handling, if it is hosted, where an untrusted image goes into a decoder — the oldest reliable route to running someone else's code on your server
- Storage cost and lifecycle, since a design tool without deletion policy accumulates every version of every draft anybody ever made
- Export determinism, where the PNG the browser renders and the PDF your exporter writes disagree about a shadow, and only one of them is going to print
Is that you?
the verdict is a default, not a law
- You are building one specific repeatable graphic — a quote card, a listing image, a thumbnail — from a template you control
- The generator is a script over your own assets, producing files into a folder, with no editor and no accounts
- Every design's source is a plain data file you could re-render with different code next year
- The fonts and stock you use are ones you can point at a licence for
- Other people will store work in it, because their designs are now hostage to your schema and your uptime
- The plan includes real-time co-editing, which is a distributed-systems project wearing a design tool's clothes
- You need a template and font library, which is a licensing exercise rather than a coding one
- It will accept image uploads from anyone you have not met
If you build it anyway
the checklist, then the prompt that enforces it
- Design the document format before the canvas, write the exporter in the same week, and version the schema from the first commit. A design tool is a format with a UI attached.
- Store designs as declarative JSON — objects, constraints, references to assets — never as serialised editor state. Editor state is a snapshot of a program that no longer exists.
- Ship export before import and before sharing: SVG or PDF plus the source JSON, so a design can leave whole rather than as a flattened picture.
- If you want anything like Smart Resize, put the layout constraints in the document from the start — anchors, minimum sizes, safe areas. It cannot be added later to coordinates.
- Keep a migration path and a test corpus of old documents. Every schema change runs against them before it ships.
- Resolve fonts explicitly and record which one was actually used in the export. Silent substitution is how a design ships in the wrong typeface.
- If it is hosted and accepts uploads, decode in a sandboxed worker, cap pixel count and byte size before decoding, sniff the real format rather than trusting the extension, and serve outputs from a domain that is not the app's.
I am building a browser design tool: photo edits, text, shapes, templates, a
brand palette, and saved designs I can reopen. The file format is what will
bite me, not the canvas. Work in this order and push back if I skip a step.
1. Define the document schema first, as declarative JSON: objects with type,
geometry, constraints and asset references, and a schemaVersion on it before
anything is saved.
2. Never serialise editor or framework state into the saved document. If a
refactor of the canvas can break my old files, the format is wrong.
3. Write the exporter in the same session as the format: SVG, PDF and the raw
JSON. I must be able to leave before I am allowed to accumulate anything.
4. Build a migration runner and a corpus of sample documents from every schema
version. Every change to the schema runs against that corpus in CI.
5. Put layout intent in the document, not coordinates: anchors, minimum and
maximum type sizes, safe margins, which objects may bleed. "Resize this
design to another aspect ratio" depends on it and cannot be retrofitted.
6. Treat the brand palette as a set of references, not copied hex values, so
changing a brand colour updates existing designs rather than orphaning them.
7. Resolve fonts explicitly. Record which font file the export actually used,
and warn visibly on substitution rather than silently reflowing my text.
8. Text layout is a real component: wrapping, minimum sizes and overflow
behaviour before templates exist. Show me a label three times too long.
9. Keep it local and single-user for as long as possible. No accounts, no cloud
storage of other people's work, no sharing links.
10. If I ask for uploads, say that an untrusted image entering a decoder is a
remote-code-execution surface, then decode in a sandboxed worker, cap pixel
count and bytes first, and sniff the real format.
11. Do not bundle a template, stock or font library. If I ask, say plainly that
embedding a typeface or a stock photo is a licensing question.
12. Out of scope: real-time collaboration, comments, permissions and team brand
management. Those are the tier I would be paying for.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
$12.99 a month for Pro, or $7.99 for Basic if you do not need the background remover, unlimited Hub storage and Brand Kit. What that money genuinely buys is the boring half: a document format somebody else maintains, a stock library licensed via Shutterstock, fonts you are permitted to use in exports, and the guarantee that a design you made in 2023 still opens. If you want the editor as a project, build the editor. If you want your work to still be editable in three years, that is the subscription's actual job.
$12.99/mo is cheaper than your weekend.
Everything depends on a decision you make in week one. If designs are JSON files plus referenced assets in a folder, the exit is free: you keep the folder, re-render with anything, and the editor's death costs you nothing. If designs live in a hosted database in a schema you invented, the exit is re-making them by hand, and that is why the data-loss tag is on this entry rather than on the photo editor next to it. Export flattened PNG or PDF copies of anything you actually shipped, as well — those stay readable regardless of what happens to the source format.
Active open-source browser design tool with an SVG-native document model — the closest reference for the part of this that is genuinely hard.
Embeddable open-source image editor, if what you want is the photo half without writing the canvas layer.
Questions
BeFunky is also DEMO ONLY. Why does PicMonkey need its own entry?
BeFunky's argument is about operations versus an editor: the filters and background removal are library calls, and the interactive canvas is the product. PicMonkey's is about persistence. Its paid tier exists to keep your designs re-editable, re-brandable and re-shapeable, which makes the document format the product rather than the canvas. Different failure, different guardrails — one is about memory and image decoders, the other is about schema migrations and export.
Is Smart Resize really that hard? It is just changing the canvas size.
Changing the canvas is trivial. Making the design still look designed afterwards is not. A square post becoming a wide banner means the headline has to shrink but not below legibility, the background has to extend rather than stretch, the logo has to keep its clear space, and the body text may have to disappear entirely because there is no room. That is a constraint solver over a document that recorded intent. If your documents only recorded x, y, width and height, the feature is not late — it is impossible.
PicMonkey is owned by Shutterstock. Does that matter for the verdict?
Only for the buy side, and it makes it stronger. The acquisition is why the stock imagery inside the editor is licensed for your use without a separate transaction, and licensed assets are consistently the part of this category that nobody rebuilds. It does not change the engineering argument at all.
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.
Every operation is a library call. The product is that it runs in a browser tab on a 40MP file without dying.
The editor is hard, the template library is harder, and the font licensing is a legal department.
You can rebuild the editor in a weekend. You cannot rebuild the licence on five million stock photos.
last reviewed 2026-08-05 · verdict is editorial and unsponsored · shared entry data from canivibecodeit under MIT · not legal advice