Vibe Coding: What It Is, the Tools and the Apps That Launched
Vibe coding is building software by telling an AI what you want and judging the app it produces rather than the code, a name Andrej Karpathy gave the practice in a post on 2 February 2025. On 19 November 2026 we read the sites of the 542 products that four launch boards listed for 1 to 19 November 2026: of the 512 that answered, 4 carried a vibe coding platform's mark in the page they served and none ran on a platform's own subdomain.
Below: the vibe coding meaning and who coined the term, those November launches as vibe coding examples, the vibe coding tools and platforms with prices from their own pages, how vibe coding compares with traditional coding, how to start, the risks as the UK's cyber security agency and the platforms describe them, and where to launch a vibe coded app once it is live.
Vibe coding means prompting an AI to build an app and accepting what it writes. You can rarely tell a vibe coded app from its page: 4 of 512 November launches carried a builder's mark, and 4 of Viberank's 145 newest apps. App builders start at $20 to $30 a month, code editors and agents at $8 to $20, and 8 of the 10 we priced keep a free plan. None of them lists the app where people browse new products: launch boards and directories do that. Viberank sells one sponsored week for $4.99; $29 at Submitator covers 40 directories you choose from the 202 in our catalog.
What is vibe coding? The meaning, and who coined it
The term comes from a post Andrej Karpathy wrote on X on 2 February 2025. It opens: “There's a new kind of coding I call "vibe coding", where you fully give in to the vibes”, and goes on to say you “forget that the code even exists”. Then the working method, in his words: “I "Accept All" always, I don't read the diffs anymore.” His own verdict on where it fits: “It's not too bad for throwaway weekend projects, but still quite amusing.”
The dictionaries now define vibe coding too. Merriam-Webster's slang and trending entry defines vibe coding as “writing computer code in a somewhat careless fashion, with AI assistance” and adds: “In vibe coding the coder does not need to understand how or why the code works, and often will have to accept that a certain number of bugs and glitches will be present.” On who coined vibe coding: “The word is believed to have been coined by Andrej Karpathy (a co-founder of OpenAI), who described the practice in a recent social media post.” Collins went further and made it its Word of the Year 2025, announced on 6 November 2025: “Coined by AI pioneer Andrej Karpathy, vibe coding refers to the use of artificial intelligence prompted by natural language to write computer code.”
Why is it called vibe coding? Because you steer by how the result feels, not by the code: in Karpathy's post you give in to the vibes, and Collins puts it as “It’s programming by vibes, not variables.” So what does vibe coding mean in practice? The vibe coding definition in the dictionaries reaches past his weekend experiment: Merriam-Webster's entry describes “the practice of writing code, making web pages, or creating apps, by just telling an AI program what you want, and letting it create the product for you.” A vibe coder is the person doing it, and a vibe coded app is the result.
Vibe coded apps: 542 November launches, read for a platform's mark
Which vibe coding apps actually launch, and can you tell them apart? Any list of the best vibe coded apps assumes you can spot which apps were vibe coded, so we checked. We took every product that Aura++, Firsto, HUNT0 and LaunchScaler listed for 1 to 19 November 2026, 542 of them, and read each one's home page as its server sends it, before any script runs. 538 linked a site and 512 of those answered. For each we asked two things: is it served on a vibe coding platform's own subdomain, and does the page carry a mark the platform leaves, such as its badge or a file from its domain?
The answer is almost never. Four of the 512 carried a mark (Lovable, Base44 and v0), on two of the four boards, and none ran on a platform subdomain. As a control we read the 150 newest apps on Viberank, a leaderboard of independent apps: 4 of the 145 that answered carried a mark (Lovable and Replit), a higher share and still under one in twenty.
Vibe coding examples, counted: 542 launches on four boards and Viberank's 150 newest apps, 19 Nov 2026
| Board | Cadence | Products listed 1 to 19 November | Sites read | On a platform subdomain | Platform mark in the page | Lead with AI |
|---|---|---|---|---|---|---|
| Aura++ | Daily | 108 | 93 | 0 | 0 of 93 | 44 of 93 |
| Firsto | Daily | 115 | 110 | 0 | 3 of 110 | 47 of 110 |
| HUNT0 | Weekly | 221 | 212 | 0 | 1 of 212 | 98 of 212 |
| LaunchScaler | Daily | 98 | 97 | 0 | 0 of 97 | 34 of 97 |
| All four boards | 542 | 512 | 0 | 4 of 512 | 223 of 512 (44%) | |
| Control: Viberank, newest 150 apps | Leaderboard | 150 | 145 | 0 | 4 of 145 | 46 of 145 (32%) |
Every product that Aura++, Firsto, HUNT0 and LaunchScaler, four launch platforms we submit to, listed with a launch date from 1 November 2026 to 19 November 2026, read on 19 November 2026 between 11:04 and 11:32 UTC, every site then read a second time from 11:56 UTC and classified again by the same rules. How each board dates a launch: Aura++, the launch stories on /launches (pages 1 to 18) plus the launches of the day on its home page, with each date taken from the launched field of the project's Markdown copy; Firsto, /launches sorted by date, pages 1 to 20, each dated by the scheduled launch day in the page's own data; HUNT0, /explore by launch week (the weeks beginning 2, 9 and 16 Nov), page after page until one came back empty; a weekly launch is dated to the Monday that opens its week, as every product page's Launch week line confirmed; LaunchScaler, /launches/archive pages 1 to 4 (November) and the day's board at /launches, each dated by its launch page. Frame: every product the four boards listed with a launch date from 1 to 19 November 2026; a product on two boards (the same host) counts once, on the first board in the order Aura++, Firsto, HUNT0, LaunchScaler; a product whose board page gives no website is listed but not read (11 listings were second listings of a product, 4 gave no website). Answered: one GET of the listed website, redirects followed, no JavaScript; answered when it returned the product's own page; a bot wall (a 403 or 429 challenge, or a page titled Just a moment, Vercel Security Checkpoint or Attention Required) leaves the base; a DNS or TLS failure, a timeout, another 4xx or 5xx, a paused deployment, a parked domain or a redirect to a page selling the domain is not answered; 5 sat behind a bot wall and 21 did not answer (nine gave no answer at all, by DNS failure or timeout, four returned a 404, three showed a paused deployment, three were parked or sent the visitor to a domain-for-sale page, one returned a 410 and one a 520). Subdomain: the listed or final host ends in a vibe coding platform's own publishing suffix: .lovable.app, .bolt.host, .replit.app or .base44.app, each read on the platform's docs; general hosting such as .vercel.app, .netlify.app, .pages.dev or .workers.dev is never a mark. Mark: the served HTML carries Lovable's badge (the lovable-badge element or its styles) or the script-tag comment found on pages Lovable hosts, loads a script, stylesheet, image, icon or meta value from a platform's own domain (lovable.dev, bolt.new, replit.com, base44.com, v0.app), or carries a generator meta tag naming a platform where that platform's own pages show the same value (v0.app, on the apps linked from v0's template gallery); links to those domains, articles that name a platform and other sites' badges are not marks; a resource on someone else's published app (a *.replit.app address, say) is not a mark either. Lead with AI: the served title or first heading names AI (as a word), artificial intelligence, GPT, LLM, an agent or agents. Control: the 150 newest app pages in Viberank's sitemap (added 2, 10 and 11 November 2026 by each page's own Added line), each app's site read by the same rules; 2 app pages gave no website and 3 sites did not answer or walled the request. No board launch is named on this page.
What a vibe coded app leaves in its page
App builders publish to an address of their own until you connect a domain: Lovable to lovable.app, Bolt to bolt.host, Replit to replit.app and Base44 to base44.app, by each one's docs. Some also leave something in the page. Lovable adds an “Edit with Lovable” badge, which its docs say Pro plans and above can remove, and pages it hosts carry a comment asking you not to remove a script tag. Bolt's free plan lists “Bolt branding on websites”. The apps linked from v0's template gallery name v0 in a generator tag. Across both sets, the marks we found were two badges, two script-tag comments, three files loaded from a platform's domain and one generator tag:
How a vibe coding platform marks the apps it builds
| Mark | What it is | Confirmed on | Board launches (512) | Viberank apps (145) |
|---|---|---|---|---|
| A platform subdomain | The app is served on the builder's own address, such as name.lovable.app, name.bolt.host, name.replit.app, name.base44.app | Each platform's publishing docs | 0 | 0 |
| Lovable's badge | The “Edit with Lovable” badge, or the styles of its lovable-badge element, left in the served page | Lovable's project settings docs, plus design-system-demo.lovable.app and hackathon.lovable.app, two apps Lovable itself hosts | 1 | 1 |
| Lovable's script-tag comment | The comment “IMPORTANT: DO NOT REMOVE THIS SCRIPT TAG OR THIS VERY COMMENT!” left in the page | build-launch-win.lovable.app, an app Lovable itself hosts | 1 | 1 |
| A file from the platform's domain | An image, script or meta value loaded from base44.com, replit.com, lovable.dev, bolt.new or v0.app | The platforms' own domains | 1 | 2 |
| v0's generator tag | A generator meta tag in the page's head, with the value “v0.app” | Seven of the eight apps linked from v0's own template gallery | 1 | 0 |
Each mark was confirmed on 19 November 2026 on the platform's own docs or on pages the platform itself hosts or links: the publishing suffixes on Lovable's, Bolt's, Replit's and Base44's docs; the badge's id and its "Edit with Lovable" text on design-system-demo.lovable.app and hackathon.lovable.app and the template comment on build-launch-win.lovable.app, all linked from Lovable's docs; v0's generator value on the deployed apps linked from its template gallery (v0.app/templates). Bolt's free plan lists "Bolt branding on websites"; no launch in either set was served from Bolt. One more board launch loads its images from a /lovable-uploads/ folder, a folder name a page Lovable hosts uses too; the rules written before the count did not list it, so it is not counted. The marks found: Lovable, Base44 and v0 on the boards, Lovable and Replit on Viberank.
Read these counts as a floor. Cursor, Claude Code and most code editors leave nothing in the page they help build, and a builder's badge comes off with a paid plan, as Lovable's and Bolt's plans say. So the census measures how many launches still show their builder, not how many were vibe coded. What it does show is that none of the 512 board launches that answered was served from a builder's address.
The control: Viberank's newest apps
Viberank's about page describes it this way: “viberank.dev is a community-driven platform that showcases and ranks the most innovative tech software and applications.” Its newest 150 app pages, added on 2, 10 and 11 November, gave 145 sites that answered. Four showed a builder (Lovable and Replit), again none on a platform subdomain. Lovable left the most marks across both sets, 4 of 8.
What the vibe coding projects were
223 of the 512 board launches (44%) put AI in their page title or first heading, against 32% on Viberank. Of the 8 apps that showed their builder, 5 led with AI, and two served a near-empty page that shows nothing until JavaScript runs, a shape 15 of the 512 board launches shared. For the jobs these products do and what they charge, our count of November launches by category and price reads the same boards from another angle.
Viberank: $4.99 for one sponsored week. Submitator: $29 for 40 directories.
Viberank is one of the 81 launch sites in our catalog, and so are the four boards above. Add 39 more picks: each one written up for that site, every badge kept live. 70 for $49, 110 for $79, one payment.
Vibe coding tools and platforms, with their prices
The AI tools for vibe coding come in two kinds. A vibe coding app builder takes a prompt in the browser, writes the app and publishes it for you: Lovable, Bolt, Replit, v0, Base44 and Google AI Studio. A code editor or coding agent works on code in your own project, on your machine or in your repository: Cursor, Claude Code, GitHub Copilot and Devin Desktop. The table lists each one's free plan and its cheapest paid plan as its own pricing page printed them on 19 Nov 2026. Lists of the best vibe coding tools rank them; this table ranks nothing, because the best vibe coding platform is the one whose limits fit what you are building.
Vibe coding platforms and AI coding tools: free plans and entry prices, 19 Nov 2026
| Tool | Kind | In its own words | Free plan | Entry paid plan | What it buys |
|---|---|---|---|---|---|
| Lovable | App builder | “Lovable is an AI software engineer, which enables anyone to build for the web.” | Yes: “No credit card needed” | Pro, $25 a month | “100 monthly credits” |
| Bolt | App builder | “Bolt is an AI app builder that turns natural language prompts into a working web or mobile app.” | Yes: “300K tokens daily limit” | Pro, $25 a month | “Start at 10M tokens per month” |
| Replit | App builder | “Describe what you want. Replit builds it. Get a working app or website in minutes. No coding required.” | No free card on the pricing page | Core, $20 a month | “$20 towards more powerful models” |
| v0 | App builder | “Build Full-Stack Web Apps with AI” | Yes: “7 message/day limit” | Plus, $30 a month per user | “$30 of included monthly credits per user” |
| Base44 | App builder | “Base44 is a no-code AI platform that uses vibe coding to build fully-functional apps and websites.” | Yes: “25 message credits” | Starter, $20 a month | “100 message credits” |
| Google AI Studio | App builder | “Turn your ideas into real apps with Gemini in Google AI Studio.” | Yes: “without a monthly subscription fee” | No subscription | “paid tiers for professional developers who need to handle massive amounts of traffic” |
| Cursor | Code editor or agent | “Built to make you extraordinarily productive, agents turn ideas into code.” | Yes: “Limited agent requests” | Plus, $8 a month | Grok models alone, by the Plus card's own note; Premium ($30) lists “Third-party models at API list price” |
| Claude Code | Code editor or agent | “Claude Code is an agentic coding tool that reads your codebase, edits files, runs commands” | No: “Claude Code is included in all paid plans.” | Pro, $20 a month | The Pro card lists “Claude Code” |
| GitHub Copilot | Code editor or agent | “GitHub Copilot works every step of the way, from an open issue to reviewing and merging a pull request.” | Yes: “2,000 completions per month” | Pro, $10 a month per user | “$15 monthly total credits for Pro” |
| Devin Desktop | Code editor or agent | “Manage any agent, locally or in the cloud, from one command center.” | Yes: “Light quota to code with agents” | Pro, $20 a month | “Full model availability” |
Each tool's own pricing page, read on 19 November 2026; the column of each tool's own words quotes its pricing page FAQ, help center, docs or home page (Claude Code's docs sentence is cut before its list ends). Prices are for monthly billing in US dollars: Lovable's pricing page printed euros when read from Europe (€25 a month, 100 monthly credits), and its plans docs give the same Pro tier as $25 in US dollars; Base44's yearly billing shows $16 a month, Replit's $18 and Claude Pro's $17. windsurf.com and its pricing page answered from devin.ai, so the last row is Devin Desktop's. Free plans carry the limits quoted; credits, tokens and messages are each maker's own unit and do not convert. The order is the table's, not a ranking, and naming a tool endorses none.
Free vibe coding tools
Eight of the 10 let you vibe code free, each with a cap. The Google AI Studio vibe coding page says you can build “without a monthly subscription fee” and deploy with one click: “Simply hit the Deploy button to host it directly on Google Cloud Run, or use the Save to GitHub option to create a permanent home for your project.” Bolt's free plan allows “1M tokens per month”; v0 gives “$5 in one-time credits for new users” and has a “7 message/day limit”; Base44's free plan carries “25 message credits” a month; GitHub Copilot's, “2,000 completions per month”. The two without one are Replit, whose pricing page starts at Core, and Claude Code, which comes with paid Claude plans.
Lovable vs Bolt, and the other builders' plans
Lovable pricing and Bolt.new pricing start at the same $25 a month and meter it differently: Lovable's Pro buys “100 monthly credits”, Bolt's Pro “Start at 10M tokens per month”. Both take their branding off on the paid plan (Bolt's Pro card says “No Bolt branding on websites”). Among the Lovable alternatives, Base44 pricing starts lower, Starter at $20 a month for “100 message credits”; Replit pricing opens at Core, $20 a month, which includes “$20 towards more powerful models”; v0 pricing at Plus, $30 per user a month. A credit, a token and a message are each maker's own unit, so compare what a plan lets you build in a month, not the numbers on the cards.
Cursor, Claude Code and Copilot for vibe coding
The editors suit someone who wants the code on their own machine. Cursor for vibe coding starts free. Cursor pricing, as its own page prints it, bundles Grok and X features into every tier: Plus costs $8 a month and, by its own note, runs Grok models alone; Premium, $30, lists “Third-party models at API list price”; and the page's FAQ says “We recommend Premium for new AI users”. Claude Code's docs open with “Claude Code is an agentic coding tool that reads your codebase, edits files, runs commands”; Claude Code for vibe coding comes with Claude Pro at $20 a month when billed monthly, so vibe coding with Claude in the terminal costs the same plan as the chat. GitHub Copilot's Pro, $10 a month, lists “Access to 3rd party agents (Claude Code and Codex)”. Replit for vibe coding is a builder in the browser, from $20 a month. Whatever you pick, the bills for hosting and AI tokens can be offset by the credit programs startups can apply for.
Vibe coding vs traditional coding
The difference is who reads the code. In traditional coding you write it and review every change; in vibe coding the AI writes it and you judge the app. The UK's National Cyber Security Centre treats it as a spectrum rather than a switch, from manual coding with autocomplete at one end to “Full vibe coding” at the other, and most work sits between:
Vibe coding vs traditional coding
| Vibe coding | Traditional coding | |
|---|---|---|
| Who writes the code | The AI, from your prompts | You, with or without autocomplete |
| What you check | The running app; Karpathy: “I "Accept All" always, I don't read the diffs anymore.” | Every change, line by line |
| Skill to start | Describing what you want clearly | A programming language and its tools |
| Where it fits | By the NCSC: “building prototypes or proof-of-concepts where speed is of the essence”, demos and internal tools | Anything, including the parts that guard accounts and data |
| When it breaks | Ask again, or paste the error back in | Read the code and debug it |
The vibe coding column quotes Andrej Karpathy's post of 2 February 2025 and the NCSC's blog "The 'vibe coding spectrum' approach to AI-assisted software development" (18 June 2026), both read on 19 November 2026; the rest describes the two ways of working and is not a measurement.
Vibe coding vs agentic coding
The two terms answer different questions. Agentic coding is about what the tool does: Claude Code's docs call it an agentic coding tool, one that reads your codebase, edits files and runs commands. Vibe coding is about what you do with the output: accept it without reading. An agent can be used either way, and the NCSC's spectrum is the dial between them.
How to start vibe coding: a path for beginners
How to vibe code well comes down to small steps and clear prompts. The seven steps below are vibe coding for beginners, a short vibe coding tutorial that works on any of the builders in the table; the order matters more than the tool:
- Pick a small first project. One user, one job, three screens. The NCSC's first example of where to slide toward full vibe coding is “building prototypes or proof-of-concepts where speed is of the essence”; demos and internal tools follow.
- Start on a free plan. Eight of the tools above have one; a builder in the browser needs nothing installed.
- Write the first prompt as a brief. Who uses it, what they do, which screens, what to leave out. A first prompt can be this plain:
Build a web app for dog walkers to book walks.
Users: dog owners and one walker.
Screens: a booking calendar, a list of my bookings, a settings page.
Owners sign in with email. The walker sees every booking.
Keep it simple and mobile first. Do not add payments yet.- Change one thing per prompt. Karpathy's example was a request to “decrease the padding on the sidebar by half”. Small asks cost less on a credit plan: Lovable's pricing FAQ prices “Make the button gray” at 0.50 credits and “Build me a landing page, use images” at 1.70.
- Keep secrets out of the browser. Lovable's docs: “Secrets stored in frontend code are visible to users and should be considered compromised.” Replit's checklist puts it as “Never expose API keys in frontend”.
- Run the builder's security check. Bolt's help center: “we recommend running a security check at least once before you publish”.
- Save the code somewhere you own. v0 lists “Sync with GitHub”, Base44 says “Two-way GitHub sync exports the full source code to your own repo at any time.” and Google AI Studio offers Save to GitHub next to its Deploy button.
Vibe coding prompts work best as briefs, not wishes: name the user, the job and the limits, and ask for one change at a time. To learn vibe coding beyond that, each builder's own docs and templates are the most current guide, since the tools change often.
Is vibe coding bad? The risks, from the sources
Not by itself, and the UK's cyber security agency says so. The NCSC's June 2026 post, “The 'vibe coding spectrum' approach to AI-assisted software development”, has a section headed “Vibe coding isn't always bad”. Its reasoning on vibe coding security starts from the training data: “AI models are trained on vast amounts of existing code, and some of that has security issues.” And its rule: “The risk isn't in using AI. The risk is not applying the right safeguards when the stakes are high”. It advises sliding toward manual work when “building authentication or authorisation logic” or “processing sensitive personal data” is in play.
The measured side of the vibe coding risks comes from code tests. Veracode, which sells code scanning, published its Fall 2026 results on 29 October 2026: “We tested 18 models across four languages (Python, C#, JavaScript, Java) and four CWE categories (CWE-89, CWE-327, CWE-80, CWE-117).” The average security pass rate rose from 56% to 62%, and the top models, it says, “still fail on more than one in four security tasks.” The builders say as much in their own docs. Lovable's security overview: “These tools help identify common security issues, but they cannot guarantee complete security.” Base44's: “That said, every app is different, and some settings depend on how your app is built and who uses it.”
So is vibe coding bad? For a prototype, no; for the code that guards logins, payments and personal data, it needs a human who reads it, and a test before it ships.
Vibe coding ideas: what to build first
The best vibe coding ideas sit where the NCSC says full vibe coding is fine and where buyers already pay: a tool for one group of people, one repeated job, little sensitive data. For vibe coding app ideas drawn from what launched, our sort of November launches by the job they do shows what charges and what does not. A vibe coding SaaS or a vibe coding startup makes money the way any small software business does, a price on the page and a way to pay, and how SaaS products structure their prices is the place to start on how to make money vibe coding.
How to launch a vibe coded app and get it found
A builder ends at publish. Everything after that, the part where strangers arrive, is yours:
- Move it to its own domain. None of the board launches in the census was still on a builder's address, and our guide to .ai, .io and .com weighs the choice.
- Check what the server sends. 15 of the 512 board launches served a page with almost no text before JavaScript ran, and on vibe coding websites it varies: two of the 8 apps that showed their builder did, the other six served their text. Open your page's source: the title, description, preview image and text there are what any reader that does not run scripts gets, and our link preview checker shows the preview your link gets on X, LinkedIn, Slack, Facebook and Google.
- Write the landing page. One promise, one call to action, proof; the sections a landing page needs are covered in our guide, and a free landing page roast scores your offer, headline, call to action and proof.
- Plan the launch. A launch plan by the week runs from the weeks before to the weeks after, and where first users come from covers the people to reach.
- List it where new products get browsed. Launch boards and directories, below.
Where to launch a vibe coded app
All four boards of the census, Aura++, Firsto, HUNT0 and LaunchScaler, are launch platforms we submit to, 4 of the 81 in our catalog. Three more in the catalog suit a newly launched app:
- Viberank (DR 52): a free leaderboard ranked by week, by month and of all time; its paid placement is a sponsored week at $4.99.
- LeanVibe (DR 28): it takes “Pre-revenue or completely free product” listings, by its home page.
- Vibecode It: its home page reads “Launch your vibe coded startup in under a minute”.
Viberank's sponsorship panel reads “$4.99 / week” and lists “Your app featured on desktop, mobile and app pages” among what it buys: one payment, one week of sponsorship on one site. A listing does two jobs a new domain cannot do for itself: it puts the app in front of people browsing that site, and it links to the app from a domain search engines already know. We run no ads and no outreach; we file listings. Our full list of startup directories shows every site, its Domain Rating and its link type.
Here is what an order involves. The picks are yours, made in a dashboard where every site shows its Domain Rating, and every site there earned its place: one of our submissions to it went through from first field to last. The filing is then ours: a description written for that directory alone, the forms and their CAPTCHAs dealt with so they never reach you. If a directory asks for a badge, a single component on your site shows whichever badges your picks require, we keep each of them live, and you never commit code for any one directory. Your listing pages are then pushed through IndexMachine by us, the free indexing ($25 value) in Starter, Growth and Scale; you set nothing up, and indexing itself stays Google's decision.
Each order is paid once: $29 buys 40 directories on Starter, $49 buys 70 on Growth and $79 buys 110 on Scale, while All Directories ($149) takes the whole catalog and the directories added later. Most directories are delivered in 24 to 48 hours and every directory within 21 days at most. If a directory turns a submission down, you pick another one by default (not on All Directories, which already holds every site), and a refund can be requested within 30 days of purchase, on the conditions in our terms. 580+ startups have been submitted so far; what became of those orders is in our submission data.
We read every source named here on 19 November 2026 with GET requests; quoted words keep their authors' spelling and punctuation. The term: Andrej Karpathy's post on X (dated 2 February 2025 by its ID, read in full), Merriam-Webster's slang and trending entry (last updated 4 Jun 2026) and Collins' blog post "Collins’ Word of the Year 2025: AI meets authenticity as society shifts" (6 November 2025). Risks: the NCSC blog "The 'vibe coding spectrum' approach to AI-assisted software development" by Toby W, Principal Security Architect (18 June 2026) and Veracode's "GenAI Code Security Fall 2026: The Ceiling Moves, But on the Low-Hanging Fruit" (29 October 2026), each read in full, with the security pages of Lovable's, Replit's, Bolt's and Base44's docs; no breach is alleged against any product. Tools: each maker's own pricing page and docs, as the table notes say; naming one endorses none. The census: 542 products from Aura++, Firsto, HUNT0 and LaunchScaler and 150 apps from Viberank, read between 11:04 and 11:32 UTC and again from 11:56 UTC by the rules under the first table. It counts marks a platform leaves in the served page, so it is a floor: editors and agents leave none, and a paid plan removes a badge. No board launch is named here. Viberank's sponsorship panel, read in a browser at 11:27 UTC, printed "Promote your app", "Your app featured on desktop, mobile and app pages" and "$4.99 / week", and the site's own scripts, read again at 12:06 UTC, carry the same price and lines; our catalog lists Viberank as a launch platform at DR 52 with a guide, so the comparison is price per payment, and this page never claims a plan of ours buys the sponsorship. The other two boards' lines are from their own home pages; their DRs and every catalog count come from our catalog export of 17 November 2026, with intercepts excluded (these 3 boards, 81 launch platforms, 202 sites in all). Our own claims trace to three sources. The owner: a description written per directory, forms and CAPTCHAs dealt with by us, a single badge component we keep live, our IndexMachine push of every listing, delivery of most directories in 24 to 48 hours and the rule that no site enters the catalog before one of our submissions to it has gone through. From the pricing cards and the home page: what each plan costs and holds, the free indexing and the 580+ figure. From the terms: the 21-day ceiling, the replacement pick when a directory says no (not on All Directories) and the 30 days in which a refund can be requested. Nothing on this page promises rankings, traffic, signups, revenue or indexing.
FAQ
Who coined the term vibe coding?
Andrej Karpathy, in a post on X on 2 February 2025: “There's a new kind of coding I call "vibe coding"”, where you “forget that the code even exists”. Merriam-Webster's entry says the word is believed to have been coined by him, and Collins named vibe coding its Word of the Year 2025.
Can I vibe code for free?
Yes, within limits. Of the 10 tools we priced on 19 Nov 2026, 8 offer a free plan or free use: Google AI Studio says it works without a monthly subscription fee, Bolt's free plan has a 300K tokens daily limit and v0 gives new users $5 in one-time credits. Replit's pricing page shows no free card, and Claude Code comes with paid Claude plans, from $20 a month.
Is vibe coding safe for a real product?
It depends on the part you are building. The UK's National Cyber Security Centre wrote in June 2026 that full vibe coding is often fine for a prototype, a demo or an internal tool, and that authentication, sensitive personal data and anything handling secrets need human review. Lovable's own docs say its security scans cannot guarantee complete security, so read and test the parts that guard accounts and data.
What is the difference between vibe coding and agentic coding?
Agentic coding describes what the tool does: an agent reads your files, edits them and runs commands on its own. Vibe coding describes what you do with the result: you judge the app, not the code, and accept changes without reading them. You can use an agent either way, reviewing every change or accepting all of them.
Can you tell if an app was vibe coded?
Rarely, from the page alone. Of 542 products launched 1 to 19 November 2026 on four launch boards, 512 sites answered: 4 carried a vibe coding platform's mark in their served HTML and none ran on a platform's subdomain. Among Viberank's newest apps it was 4 of 145. Code editors and agents leave nothing, and Lovable and Bolt let paying users remove their badge or branding.
Is vibe coding the future of software?
For some of it, by the reading of the UK's National Cyber Security Centre, which wrote: “It's possible that we'll trust vibe coding more overtime as models become more reliable and their outputs become more trustworthy.” It still advises human review where a flaw would be costly, and Veracode's tests of 18 AI models, published in October 2026, put the average security pass rate at 62%.
Where should I launch a vibe coded app?
Where people browse new products: launch boards and directories. Viberank is a free leaderboard of independent apps and LeanVibe takes pre-revenue or completely free products; both are in our catalog, with the four general boards of this census. Give the app its own domain and a landing page first.
Keep reading
Their $4.99: one sponsored week. Our $29: 40 directories.
Viberank charges $4.99 for one week of sponsorship on one leaderboard. For $29 Submitator files 40 directories of your choosing, or 110 for $79, paid once: a description written for each site, and the badges kept live.
