Should I vibe code
Capture a browser workflow and produce editable step-by-step documentation
Auto-generating step-by-step docs is capture plus OCR plus restraint.
?
Their verdict, the Pro Personal 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
The pipeline is buildable and the docs are genuinely handy. Same caveat as any screen capture tool: redact before you publish, because it captures whatever was on screen.
What actually breaks
not "if". the specific failures.
- Redaction, because the screenshots are of a real screen with real customer records on it
- Automatic blurring, which finds the fields it was told about and misses the customer name in the page title
- The sidebar, the notification toast, the browser tab titles — everything captured incidentally around the thing you meant to document
- Step detection, which has to infer intent from clicks and produces twelve steps for what a human would call three
- Documents that go stale the moment the UI changes, with no signal that they have
You document the refund process using a real order, because that is the only way to see the real screens. Your blur rule covers the email field. It does not cover the browser tab, which reads 'Order #4471 — Marguerite Dubois', or the notification that slid in at step four showing another customer's message. The guide gets published to the help centre. It is genuinely useful, it is shared widely, and it contains two people's names that nobody ever decided to publish.
Is that you?
the verdict is a default, not a law
- Every capture is on a dedicated demo account with fabricated data
- The output stays internal and is reviewed before anyone else sees it
- You redact by hand, deliberately, on every screenshot
- Captures are taken against production with real customer data
- Redaction is automatic and unreviewed
- Documents are published externally without a human checking every frame
- Nothing tells you when a documented flow has changed
If you build it anyway
the checklist, then the prompt that enforces it
- Capture from a demo account with fabricated data. This removes the entire redaction problem and is far more reliable than blurring.
- If you must capture production, require a manual review of every frame before export, and make review the default rather than a setting.
- Redact by allowlist, not blocklist: blur everything, then reveal the region being demonstrated. Blocklists miss tab titles and toasts.
- Capture the window rather than the screen, and suppress notifications during recording.
- Store the raw capture separately from the published version so a mistaken redaction can be corrected rather than re-recorded.
- Stamp each document with the date and the app version it depicts, so staleness is visible.
Before you build a screen-capture documentation tool, apply these and push back if I ask you to break them. 1. Ask me first whether captures will be taken against production data. If yes, tell me to create a demo account with fabricated records instead, and explain that this eliminates the redaction problem rather than mitigating it. 2. If production capture is unavoidable, implement redaction as an allowlist: blur the entire frame by default and reveal only the region being demonstrated. Explain that blocklist redaction reliably misses browser tab titles, notification toasts, sidebars and breadcrumbs. 3. Require human review of every captured frame before export, and make review mandatory rather than a preference. Do not build a publish-without-review path. 4. Capture a specific window rather than the whole screen, and suppress system notifications while recording. 5. Keep the original capture and the redacted output as separate artefacts, so a bad redaction can be fixed without re-recording the workflow. 6. Group related clicks into meaningful steps and let the user merge and rename them. Raw click capture produces twelve steps where a human means three. 7. Record the date and application version on every document, and surface documents older than a threshold as possibly stale. 8. Strip metadata from exported images. 9. Out of scope unless I ask: video export, voiceover, branching guides, embedded analytics.
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
$29 a month is reasonable if you document customer-facing processes regularly, mostly because their redaction and review flow has already met the tab-title problem. If your captures are all on demo data and stay internal, building this is a good weekend with a much smaller risk surface.
$29/mo is cheaper than your weekend.
Export guides as Markdown plus a folder of images, so the documentation outlives the tool that made it. Keep originals and published versions separate, and keep the demo account — being able to re-record cleanly is what stops a UI change turning into a redaction review.
Open-source screen recording and sharing product with modern desktop architecture.
Questions
Why is allowlist redaction better than blurring known fields?
Because you cannot enumerate everything that might carry data. A blocklist covers the fields you thought of — email, phone, address — and misses the tab title, the breadcrumb, the toast that appeared for two seconds, the autocomplete dropdown. Blurring everything and revealing only what you are demonstrating fails safe instead.
Isn't a demo account a lot of setup?
It is an afternoon, once, and it removes the hardest ongoing problem entirely. It also makes documentation better — fabricated data can be chosen to illustrate the point clearly, rather than whatever real record happened to be open when you hit record.
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.
Automatic cursor zoom looks like magic because getting it to feel right is genuinely hard.
Capture, annotate, copy. Every piece of this is a platform API you already have.
Screen plus camera to a file is platform APIs. The polish is where the money went.
last reviewed 2026-08-03 · verdict is editorial and unsponsored · shared entry data from canivibecodeit under MIT · not legal advice