Bolt.new Clone App - White-Label AI Code Generation Platform
This is a ready-made, white-label bolt.new clone you host yourself: the source code for a browser workspace that turns a plain-language prompt into a running multi-file project, plus an operator console and desktop builds, rebranded and live within six working days.
Users choose a model from 22 LLM providers, edit in CodeMirror beside a terminal, preview in the browser and deploy, while you set credit rates per model and plan. There is no mobile app, and the StackBlitz WebContainers license for commercial use is one the operator obtains. Adding a provider or a plan rule of your own is custom work: we price it in writing, and most such builds take 2-8 weeks.
Launch in 6 Days with
- Source Code
- Web Workspace
- Desktop Builds
- Admin Console
- Docker Setup
- Your Branding
StackRemix · React · Electron · WebContainers · Express · workerd · PostgreSQL
Activity as it happens
- App generated from promptGeneration
- Credits written to ledgerMetering
- Request routed to providerProviders
- Token ceiling reachedGovernance
- Project deployed to NetlifyDeploy
- Bonus credits grantedAdmin
- Plan changed to ProUsers
- ZIP export downloadedExport
- App generated from promptGeneration
- Credits written to ledgerMetering
- Request routed to providerProviders
- Token ceiling reachedGovernance
- Project deployed to NetlifyDeploy
- Bonus credits grantedAdmin
- Plan changed to ProUsers
- ZIP export downloadedExport

- 6Money-In Routesplan memberships, per-model credits, key margin, seat licensing, promotional credits and white-label resale
- 6Deployment TargetsVercel, Netlify, GitHub, GitLab, Cloudflare Pages and ZIP
- 0%Revenue Share With Miracuvescredit and plan income is yours, with no per-seat charge
- 3Desktop SystemsmacOS, Windows and Linux
- 6Days to Launchand if the delay is ours, we keep going free
- $3,699Paid Once, Not Rentedsource code yours, no per-seat fee, no monthly rental
Live in six working days, or we keep working free until you launch. Every build carries 60 days guidance, 6 months bug fixes and 12 months updates.Verify every figure
Shipped for a Team Under NDA
An enterprise platform team in India, name withheld under NDA, launched a self-hosted AI build platform on this base in 5 weeks, with 22 providers unified.
Keys Stay Server-Side
The Express auth proxy injects your provider keys server-side, and every credit change by an admin is written to the ledger with a mandatory reason.
Payments Wire In at Deployment
The platform meters usage in credits but processes no payments; Stripe or another provider is connected at deployment as an add-on, with webhook signatures verified.
01See it live
Live in Action
Bolt.new Clone Demo - Builder Workspace, Pro, Enterprise & Oversight Panel
You do not have to rely on our description - try the Miracuves Bolt.new Clone for yourself. Four logins work across every tier and nothing needs setting up. Create an app from a prompt, see it run in the browser, then log in as the operator and adjust what each tier may use.

Bolt.new Clone - Free Tier
The core build loop - describe an app in plain language, watch multi-file code generate, run it live in the browser, then edit in CodeMirror with a real terminal beside it.
- 1Open the web application in whichever browser you prefer
- 2Use the demo account details below to sign in
- 3Explore: Chat, Editor, Terminal, Preview
demo@mxbolt.com | demo1234
All 22 Providers & Deployment
The paid build surface - every provider and model unlocked, the full template library, prompt enhancement, custom system prompts, MCP tools and one-click deployment.
- 1Open the web application in whichever browser you prefer
- 2Use the demo account details below to sign in
- 3Explore: Providers, Templates, Deploy, MCP
pro@mxbolt.com | pro1234
Unlimited Credits & Team Scope
The enterprise surface - credit metering bypassed entirely, extended context windows, white-label branding, custom agents and programmatic API access.
- 1Open the web application in whichever browser you prefer
- 2Use the demo account details below to sign in
- 3Explore: Unlimited usage, Branding, API
enterprise@mxbolt.com | ent1234
Operator Control Plane
Where the platform is actually run - provider keys, per-plan model and pricing configuration, user and credit administration, feature flags and environment variables.
- 1Open the admin panel in your browser
- 2Use the demo account details below to sign in
- 3Explore: Keys, Plans, Users, Flags
admin@mxbolt.com | admin1234See It Working
Bolt.new Clone Video Demo: Generation, Preview & Operator Console Walkthrough
Want a guided tour? Book a 30-minute call with our team and map your launch – provider strategy, credit pricing, plan design, deployment targets and go-live sequencing.
Schedule a live walkthrough call.Every View Documented
Bolt.new Clone App Flows - Builder, Deployment, Admin & Integration Screens
Understand exactly how each part of the platform works. The build loop, the deployment path, the operator console and the integration layer are separate surfaces with distinct users – walk through each before you commit.


















02Proof
Client Voices
What Miracuves Clients Say - Actual Platforms, Recorded Results
Feedback from operators who launched their platforms with Miracuves.
“Multi-tenant structure, alert pipelines and agent collectors were the slow part and they already existed. We added our scoring engine, the playbook library and billing. First customers were onboarded within weeks and the dashboard is now a sales tool rather than an internal screen.”
“GUS is a B2B utility-billing SaaS for small operators across LATAM — meter reads, billing runs, dunning, and a customer portal. The billing engine is the most failure-prone part of the whole stack, and we didn’t want to maintain it ourselves. Miracuves’s MXBilling base gave us the billing engine, the dunning workflow, and the customer portal. We added our meter-ingestion adapters, our regional tax rules, and a custom multi-tenant reporting layer. We went from kickoff to our first operator live in 11 weeks. The platform has been running for two billing cycles with no major incidents — exactly the reliability profile we needed.”
“Conversational workflows, knowledge-base integration and the automation layer were already built. We connected our own data sources and tuned multilingual accuracy for support and lead qualification. The team now runs the whole AI operation from one place.”
“We needed seven Web3 products off one spine rather than five engineering teams. Shared auth, KYC and wallet came from their suite. Each product landed in weeks rather than quarters, and our ops cost per product is a fraction of a stand-alone build.”
Proof in Production
Bolt.new Clone Case Study - Self-Hosted AI Build Platform Deployment
How an enterprise platform team stood up a governed AI code generation environment on the Miracuves Bolt.new Clone. Client identity withheld under NDA.
A self-hosted AI code generation platform deployed on the Miracuves Bolt.new Clone, with shared provider keys and per-team credit governance from day one.

- Giving engineers AI tooling without sending proprietary code to a third-party SaaS
- Central provider keys so individual teams never handle raw API credentials
- Per-team credit metering that made AI spend forecastable instead of a surprise
- Stand up an AI code generation platform without building auth, billing and governance from zero
- Give the platform team real controls over models, cost and access
- Make AI spend attributable to a team and a model
- Browser-native execution with no server-side sandbox
- Credit metering, plan tiers & shared provider keys
- Operator-configurable models, pricing and feature flags
- Generate, edit, preview, deploy & export in one workspace
- Express auth proxy with a Cloudflare workerd runtime
“The admin panel is what got this past our security review. Central keys, per-plan models, and nothing leaving our network.”Head of Platform Engineering · Enterprise platform team - name withheld under NDA
03The product
The Basics
What Is a Bolt.new Clone App?
A Bolt.new Clone App is a ready-made AI code generation platform that recreates the prompt-to-running-app experience: a user describes software in natural language, the AI writes a complete multi-file project, and that project boots and runs immediately inside the browser. Around that core sit the things a commercial deployment actually needs – accounts and sessions, subscription tiers, credit metering per model, an operator console for providers and pricing, and one-click deployment to the platforms your users already ship on. You own the source, run it on your own infrastructure, and set your own commercial terms.

Built for Web, Desktop & API
Browser-first, with packaged Electron builds for macOS, Windows and Linux, and programmatic API access on the enterprise tier.
White-Label & Rebrandable
Your name, colors, theme and domain. Enterprise-tier white-label branding is a first-class feature, not a find-and-replace exercise.
Governed From Day One
Roles, plans, credit metering, provider permissions and feature flags are in the product already - not a phase-two project.
The Engine
What Is a Bolt.new Clone Script?
A Bolt.new Clone Script is the underlying source code and architecture that makes that possible – the Remix application, the Express authentication proxy, the Cloudflare workerd runtime, the PostgreSQL schema and the provider abstraction layer that turns twenty-two different LLM APIs into one interface.

With the Miracuves Bolt.new clone script, you:
- 01Own 100% of the source code
- 02Launch in 6 days instead of waiting months
- 03Avoid long-term vendor lock-in
- 04Add custom providers, models and workflows over time
In simple words: it is a shortcut to launching an AI app-builder product without spending a year building the generation pipeline, the provider routing, the billing model and the governance layer that make it sellable.
Everything Included
Bolt.new Clone Features - Generation, Providers, Workbench, Deployment & Governance
The platform covers the full path from prompt to deployed application, plus the commercial layer around it. Readiness is stated per capability using the same language as the technical documentation – Included, Supported, or Configuration required – so nothing surprises you after purchase.
Natural Language to Running App
Describe it, and the AI writes the full project, then boots it in the browser.
- Complete multi-file projects from a single description
- Frontend, backend routing, config and dependency manifests
- Streaming responses with multi-segment continuation
Multi-File Project Awareness
The AI reads and modifies across the whole file tree at once.
- Reads and modifies across the whole file tree at once
- Context selection against the model token budget
- Older turns truncated before system prompt or recent messages
WebContainers Runtime
A complete Node.js environment executing inside the user’s browser.
- A real Node.js environment running client-side
- npm install and shell commands in the browser
- Boots in three to five seconds, no remote sandbox
Live Preview With Hot Reload
File writes reach the preview pane in under half a second.
- Hot reload into the preview pane in under half a second
- No build step, no deploy, no waiting
- Framework-aware for ten frameworks
CodeMirror 6 Editor & Diff View
Every AI change is reviewable before anything lands in your project.
- Multi-file tabs with syntax highlighting
- Visual diff of every AI change before you accept it
- Ten-plus language modes supported
Terminal With Error Feedback
A real shell whose failures can be returned to the AI for correction.
- A real shell inside the container
- Multiple terminals on paid tiers
- Failed commands returned to the AI for correction
22 LLM Providers
One unified selector across every major provider, plus local models via Ollama.
- 22 providers behind one BaseProvider contract
- Four on Free, all twenty-two on Pro and Enterprise
- Documented four-step key resolution chain
OpenRouter Model Access
Hundreds of additional models, free and paid, from a searchable catalog.
- 365+ additional models through OpenRouter
- Each tagged free or paid with context length
- Catalog cached five minutes with static fallback
Template Library
Starter projects across eight categories, gated per plan by the operator.
- 30+ starter projects across eight categories
- Landing pages, web apps, dashboards, e-commerce
- Social, games, AI tools and education
Prompt & Context Tooling
Enhancement, custom system prompts and budget-aware context selection.
- Prompt enhancement rewrites a vague brief
- Custom system prompts on paid tiers
- Per-project AI behavior configuration
Six Deployment Targets
Ship to the platforms your users already use, or export the project as a ZIP.
- Vercel, Netlify, GitHub, GitLab, Cloudflare Pages, ZIP
- Framework auto-detection across ten frameworks
- Deployment status polling with the live URL surfaced
Git, Supabase & MCP Integration
Import repositories, generate database schemas, and extend the AI with external tools.
- GitHub and GitLab import through an SSRF-protected proxy
- Supabase schema, RLS policies and client code
- MCP tools over stdio, SSE and streamable-HTTP
Every capability above is white-label and operator-configurable through 45+ feature flags across eight categories. Deployment targets, Supabase and MCP require the user’s own account or endpoint – the documentation marks these Integration required, and so do we.
Where Operators Make Money
Bolt.new Clone Revenue Models - Subscriptions, Credits, Seats & Licensing
The commercial model is built into the schema rather than bolted on. Three subscription tiers, per-model credit pricing you control, and an enterprise tier that bypasses metering for organizations that account for AI spend centrally.
- 01
Plan-Based Memberships
Free, Pro and Enterprise with distinct provider access, credit allocations, template libraries and deployment permissions.
- 02
Per-Model Credit Pricing
Set input and output credit rates for every provider and model on every plan. A cheap open-source model and a frontier model can price honestly against each other.
- 03
Margin on Shared Keys
You buy provider capacity at wholesale and meter it to users in credits. The spread between your API cost and your credit price is the business.
- 04
Enterprise Seat Licensing
Unlimited-credit accounts with seat allocation, for organizations that would rather pay a flat platform fee than meter individuals.
- 05
Bonus & Promotional Credits
Admins grant credit blocks for trials, onboarding or retention, each written to the transaction ledger with a stated reason.
- 06
White-Label Reselling
Enterprise-tier branding lets agencies and platform vendors deploy the product under their own name for their own customers.
| Revenue stream | How it is charged | Take rateTypical range | What sets the ceiling |
|---|---|---|---|
| Plan-Based Memberships | A flat monthly price per tier, with the token allowance and feature set rising at each step. | Bolt Pro $18 a month, Teams $27 per member (annual billing); Replit Core $20. | Competitor anchors. Lovable's Pro starts at $25 a month for 100 credits and Replit Pro is $100, so builder subscriptions sit in a band of roughly $18 to $100 before usage. |
| Per-Model Credit Pricing | Each provider model gets its own input and output credit rate, scaled to what the provider charges per token. | Provider lists run $0.10 to $10 per million input tokens, $0.50 to $50 output. | Provider price sheets. OpenAI and Anthropic both list an entry model at $0.10 input and $0.50 output and a flagship at $10 and $50 per million, a 100x range. |
| Margin on Shared Keys | You pay the provider's per-token list price, then charge users credits that cost them more than that. | $18 per 10M tokens on Bolt Pro, about $1.80 per million (arithmetic on published figures). | Bolt's token is its own meter, so the comparison is not like for like. Anthropic lists Sonnet 5.5 at $2 input and $0.10 for cache hits, so caching and model mix drive margin as much as markup. |
| Enterprise Seat Licensing | A per-member monthly fee with a larger pooled token allowance, billed annually for teams. | Bolt Teams: $27 per member a month, starting at 10M tokens each. | Team budgets. Lovable Business starts at $50 a month for 100 credits per workspace and Replit Pro is $100 for up to 15 collaborators, so seat pricing is read against whole-team tool spend. |
How to read this. Monthly prices and token allowances were copied from the Bolt, Replit and Lovable pricing pages. Provider costs come from the public price lists of OpenAI and Anthropic. The $1.80 per million figure is our own division of Bolt's published price by its published allowance, not a rate Bolt states. Plans change often, so recheck before you lock a price. Everything here is observed market pricing with no regulatory standing, and it is no guide to how much your platform brings in.
Important: The platform meters, prices and ledgers usage – it does not process payments. There is no built-in checkout. Connecting a payment provider such as Stripe is operator work at deployment, and the documentation states this plainly.
Talk to Our TeamOperate It With No Developers on Staff
Bolt.new Clone Admin Console - Providers, Pricing, Users & Feature Toggles
The operator console is the reason this is a product rather than a demo. Everything that determines cost, access and behavior is configurable at runtime by an authenticated admin – no redeploy, no code change.


Management Sections
- 01
Live Dashboard
Total users, active today, new signups this week, plan distribution and credits consumed.
- 02
Provider Key Management
Create, update, rotate and deactivate API keys across all 22 providers, with masked previews.
- 03
Plan-Provider Configuration
Assign models from any provider to any plan, with per-model input and output credit rates.
- 04
Per-Request Token Ceilings
Cap context size per plan and model, independently of the model’s native window.
- 05
User Administration
Search by email, username or plan, and change any user’s tier from one screen.
- 06
Credit Adjustments
Add or deduct credits with a mandatory reason, written to the transaction ledger.
- 07
Account Suspension
Deactivate an account instantly without deleting its history or transaction record.
- 08
45+ Feature Flags
Eight categories, validated server-side - a modified cookie changes nothing.
- 09
Environment Variables
Read and set runtime configuration with sensitive values masked in the interface.
- 10
Template Library Control
Manage templates, categories, preview covers and which plans may use them.
- Platform statistics: total users, active today, new signups this week
- Plan distribution across Free, Pro and Enterprise
- Credits consumed and purchased, with per-provider request volume and top models
- Environment variables readable and settable with secrets masked in the UI
- Template library management including categories, covers and per-plan gating
- Subscription records with status, period and cancellation state
Who gets in: Admin authentication uses a separate session with its own cookie and a shorter 24-hour expiry, validated with a timing-safe comparison. If no admin password is configured, admin login fails outright rather than falling back to a default.
Talk to Our Team04Your deliverables and the price tag
The Full Package
What's Included - Bolt.new Clone Source Code, App Builds & Hosting
You receive the complete platform and its documentation set, not a stripped demo build.

Web Application
The complete Remix and React build-and-preview workspace, on your own domain.
- Remix 2.15 and React 18.3 with UnoCSS styling
- CodeMirror 6 editor and xterm.js terminal
- 25 nanostores with localStorage persistence
$ electron · auto-update Desktop Builds
Electron packaging for macOS, Windows and Linux, with auto-update.
- Electron 33 packaging for macOS, Windows and Linux
- Auto-update support and native logging
- Built from the same codebase as the web app
$ chat request → proxy Authentication Proxy
The Express service owning sessions, credits and provider key injection.
- Express 5.2 authentication proxy
- Sessions, rate limiting and database operations
- Server-side provider key injection

AI Runtime
The workerd process carrying generation, streaming, MCP and deployment.
- Cloudflare workerd AI runtime
- LLM streaming with recovery and MCP orchestration
- Deployment and git proxying
$ postgres.js · 6 tables Database & Provider Layer
The full PostgreSQL schema and the abstraction covering all 22 providers.
- PostgreSQL schema across six tables
- Provider abstraction covering all 22 providers
- Parameterized queries throughout, no ORM
API & Integration Layer
REST endpoints across eleven modules plus every third-party integration service.
- REST endpoints across eleven modules
- GitHub, GitLab, Vercel, Netlify and Supabase integration code
- MCP client over three transports
Branding & Re-Skinning
Your name, logo, colors, theme and domain applied throughout.
- Your name, logo, colors, theme and domain
- Enterprise white-label branding included
- No visible Miracuves branding anywhere in the product
Deployment & Support
Cloud setup, Docker and Kubernetes configs, plus post-launch support and updates.
- Cloud setup on your preferred provider
- Docker Compose and Kubernetes manifests included
- 60 days platform guidance, 6 months priority bug fixes, 1 year of updates
Scope note: There is no mobile application. The platform is browser-first with packaged desktop builds – that is what the product is, and we would rather say so than let you discover it later. Native mobile is available as custom development if your roadmap requires it.
Talk to Our TeamPlainly Priced
How Much Does It Cost to Build an AI App Builder Like Bolt.new in 2026?
Building an AI code generation platform with Miracuves typically starts from around $3,699 for a launch-ready white-label Bolt.new Clone and increases based on provider strategy, credit pricing design, deployment targets, integrations and infrastructure complexity.
Professional
ReadyMade Launch BundleStrongest fit: Dev-tool startups, agencies, AI product teams & internal platform groups
What you receive is the same product as the demo, wearing your brand and running on infrastructure you control.
Get StartedEnterprise
Tailored, Regulated & Big-ScaleStrongest fit: Enterprises, universities, regulated engineering orgs & white-label operators
Or have it built to fit your circumstances and business requirements, tailored to whatever you specify.
Request a Quote05Who it is for
Know Your Buyer
Who Is Our Bolt.new Clone App For?
This suits anyone whose business depends on turning descriptions into working software – whether you sell that capability, or supply it internally.
Shrink client prototyping from weeks to days while every deliverable stays under your agency's brand.
- Walk clients through a running preview that hot-reloads as changes land
- Hold provider keys centrally so no team member handles raw API credentials
- Deliver through Vercel, Netlify, GitHub, GitLab, Cloudflare Pages or a ZIP archive
Offer AI-assisted development across the engineering organization without sending proprietary code to an outside SaaS.
- Generated code executes in each engineer's browser, so your servers never run it
- Attribute AI spend to a team and a model through per-plan credit metering
- Hand reviewers a VAPT report alongside the developer security handbook
Give each student the same zero-install coding environment, whatever laptop they bring to class.
- Node.js, npm installs and a real terminal running client-side, with nothing to set up locally
- Thirty-plus starter templates, plus your own, gated per plan from the operator console
- Credit allocations and token ceilings that keep classroom AI usage within budget
Embed an app builder inside your own product, or run several branded builders for different markets from one codebase.
- Rebrand name, logo, palette, theme and domain with no Miracuves branding visible
- Consultancies can bundle the platform into delivery alongside their own services
- Own the source outright, with no revenue share or per-seat fee owed to Miracuves
Bring a branded AI app builder to market on the Miracuves Bolt.new Clone, then sell it as a metered product on your own terms.
- Users turn plain-language briefs into complete multi-file projects that run in their browser
- Free, Pro and Enterprise tiers with input and output credit rates you set per model
- Twenty-two LLM providers plus 365+ OpenRouter models, switched on or off per plan
Use the generation pipeline as the engine underneath a product aimed at one industry or one kind of user.
- Seed every session with your own starter templates and custom system prompts
- Add providers or local Ollama models as configuration, not as a rebuild
- Skip a year of work on routing, billing logic and governance before your first release
If you need AI-assisted development with control over models, cost and where code executes, this is the shape of product you are looking for.
Talk to Our TeamFit by Setting
Bolt.new Clone Use Cases - Dev Tools, Agencies, Enterprise Platforms & Education
The same codebase fits very different operators, depending on which layer they lean on. Commercial products focus on credit metering and tiers; internal platforms focus on governance and unlimited seats; education deployments focus on templates and cost control. Each case is handled through configuration rather than a fork.
Swipe · 6
Sell AI app generation as a metered product, with credit pricing set per model and per plan so your margin tracks provider economics instead of guessing at them.

Compress client prototyping from weeks to days, deploying under your own brand with central provider keys so no individual account handles raw API credentials.

Supply AI-assisted development across an engineering organization, where browser-side execution and self-hosting are what satisfy the security review rather than a policy exception.

Give every student a consistent zero-install environment, with faculty-authored assignment templates, per-plan gating and ZIP export for submission.

Use the generation pipeline as the layer beneath a vertical AI product, swapping models and pricing as the market moves without touching the application above it.

Deploy repeatedly under different brands for different markets, with white-label branding, per-deployment provider strategy and independent commercial terms.

One platform, many variants. Every use case above runs on the same schema and the same four roles. What changes is the plan structure, the credit pricing, the provider mix and the branding – not a fork you then maintain separately.
Talk to Our TeamMarket Timing
Why Launch an AI Code Generation Platform in 2026?
AI-assisted development stopped being a novelty and became infrastructure. The unresolved questions are commercial and operational – which models, at what cost, running where, and who can use them. Those are exactly the questions this platform answers, and exactly what the hosted alternatives will not let you control.
01Own the Tooling, Not Rent It
Hosted AI coding tools price per seat and bind you to their model choices. Self-hosting makes cost a function of usage and keeps model selection yours.
02Code Never Leaves the Browser
Generation happens in the AI pipeline; execution happens client-side in WebContainers. For teams with IP or compliance constraints, this is the deciding argument.
03Multi-Provider Is the Hedge
Model quality, pricing and availability move constantly. Twenty-two providers behind one interface means a pricing change is a config edit, not a migration.
04Zero Preview Infrastructure
Competitors run remote VMs or containers for every preview session. Browser-native execution removes that cost line entirely.
05Sell It, Or Supply It
The same build works as a commercial product with metered tiers, or an internal platform with unlimited enterprise accounts and central governance.
06Governance Is the Differentiator
Provider keys, per-plan model access, credit ceilings and feature flags are what turn an AI demo into something procurement will actually approve.
By the Figures
LLM Providers
Starter Templates
Deploy Targets
Feature Switches
The window is open because the incumbents optimized for individual developers, not for the organizations that have to pay for them, govern them and pass a security review with them.
Industries
Industries We Serve
The Bolt.new Clone suits any organization that benefits from turning descriptions into working software under its own control. Developer-tool companies sell it directly as a metered product. Agencies and studios use it to compress client prototyping from weeks to days while keeping the work under their own brand. Enterprise platform teams supply AI-assisted development internally, where browser-side execution is what satisfies the security review. Universities and training providers give every student a consistent zero-install environment. AI product startups use it as the generation layer beneath a vertical product. Consultancies bundle it into delivery. SaaS vendors embed it as an in-product app builder for their own customers. White-label operators deploy it repeatedly under different brands for different markets.
one platformMXBolt
- 01DtDeveloper Tools
- 02EpEnterprise Platform Teams
- 03AsAgencies & Studios
- 04EtEducation & Training
- 05ApAI Product Startups
- 06SvSaaS Vendors
- 07CoConsultancies
- 08RdRegional Dev Platforms
- 09RlR&D & Innovation Labs
- 10ReRegulated Engineering
- 11AiAccelerators & Incubators
- 12WlOperators Under Your Brand
Miracuves’ Bolt.new Clone is built as a governance-first AI code generation platform adaptable to any commercial model – metered where your business requires it, and white-labeled entirely under your brand.
06Under the hood
Across the Full Journey
How the Bolt.new Clone Works - Prompt, Generate, Preview, Iterate & Deploy
An AI builder looks simple from the outside and runs on a tightly sequenced identity, generation, execution and metering pipeline underneath. Here is the path from registration to a deployed application and a billed account.
User Journey - From Signup to Deployed App
Swipe · 6
- 1Signup & Provisioning
A user registers and the platform provisions their account, tier and starting balance before they touch the workbench.
- Email, username and password validated at the API boundary
- Password hashed with bcrypt at 12 salt rounds
- Session token generated with 256 bits of entropy, stored as a SHA-256 hash
- Free tier assigned with a welcome credit balance
- 2Choosing a Starting Point
They pick a template or start blank. The template seeds the browser runtime with a complete, working project rather than an empty folder.
- 30+ templates across eight categories, each with its own generation prompt
- Category filtering plus search across name, description and tags
- Per-plan template gating configured by the operator
- The chosen template seeds the WebContainer with a full project structure
- 3Prompt & Generation
They describe what they want. The request is routed, keyed, metered and dispatched to the selected model.
- Conversation history, file state and model configuration serialized
- The Express proxy injects operator provider keys for eligible plans
- workerd invokes the selected model through the Vercel AI SDK
- Response streams back as interleaved text and action commands
- 4Execution & Live Preview
The action commands execute inside the browser. Nothing runs on your servers, and the preview updates as the files land.
- File writes, shell commands and dependency installs run in the WebContainer
- Changes hot-reload into the preview pane in under half a second
- Every change reviewable as a visual diff before acceptance
- Terminal output available for any command that fails
- 5Iteration & Metering
They refine through follow-up prompts or direct edits, and the platform accounts for what was actually consumed.
- Full multi-file awareness carried across follow-up turns
- Failed terminal commands can be returned to the AI for correction
- Cost computed from actual input and output token counts
- Balance validated server-side, then written to an append-only ledger
- 6Deployment & Export
They ship it - to a hosting platform, a git remote, or straight to disk.
- Vercel, Netlify, GitHub, GitLab, Cloudflare Pages or ZIP export
- Framework auto-detected across ten frameworks
- Deployment status polled and the live URL surfaced in the workbench
- Full project archive downloadable at any point
How It's Built
Bolt.new Clone Platform Architecture & Back-End Path
- 01Layer 01
Request Routing
The Express proxy intercepts auth, credit, admin and provider routes; everything else forwards to workerd. One public entry point, two runtimes.
- 02Layer 02
Provider Key Injection
On a chat request the proxy resolves the user’s plan, fills any provider gap with an operator key, rewrites the header and forwards. The browser never receives it.
- 03Layer 03
Credit Metering
Cost is computed from actual input and output token counts against the plan-model rate, deducted server-side, and written to an append-only ledger with the balance after.
- 04Layer 04
Streaming & Recovery
A switchable stream aggregates multi-step responses with cumulative usage tracking, wrapped in a 45-second timeout and two automatic retries.
- 05Layer 05
Feature Gating
Flags resolve from a cookie set at session validation, then re-validate server-side in workerd - under a millisecond, with no database round trip and no client trust.
- 06Layer 06
Browser Execution
Generated code never runs on your servers. WebContainers execute it client-side, which is both the cost argument and the compliance argument.
Under the Hood
Bolt.new Clone Tech Stack - Remix, workerd, PostgreSQL & WebContainers
The architecture is deliberately asymmetric: an Express proxy owns everything that touches the database, and a Cloudflare workerd runtime owns everything that touches the AI pipeline. The workerd runtime is fast and edge-friendly but cannot open raw TCP connections, which PostgreSQL requires – so rather than compromise on either, the Express proxy acts as the database bridge and forwards everything else through. That produces three concrete properties: database credentials exist only in the Express process and workerd can never expose them, the two tiers scale independently, and if the database goes down authentication degrades to 503 while the AI pipeline keeps working for users on their own keys.
Frontend
Remix 2.15 · React 18.3 · UnoCSS · CodeMirror 6 · xterm.js · nanostores- Server-rendered Remix application with Vite and hot module replacement
- 25 nanostores with localStorage persistence and cross-tab synchronization
- CodeMirror 6 editor and an xterm.js terminal in an integrated workbench
Browser Runtime
StackBlitz WebContainers- A complete Node.js environment executing inside the browser
- npm install, shell commands and a real filesystem API, client-side
- Boots in three to five seconds; preview hot-reload under half a second
Auth & Data Layer
Express 5.2 · PostgreSQL · postgres.js- Session management, rate limiting, credit metering and provider key injection
- Six tables accessed through parameterized tagged-template queries, no ORM
- Connection pooling capped at ten with idle and connect timeouts
AI Pipeline
Cloudflare workerd · Vercel AI SDK 4.3- Provider abstraction over 22 LLM SDKs behind one BaseProvider contract
- Streaming with a 45-second timeout, two retries and structured error annotation
- Context selection against the model token budget before each call
Integrations
GitHub · GitLab · Vercel · Netlify · Supabase · MCP- Repository import and push through an SSRF-protected proxy
- Deployment with framework auto-detection across ten frameworks
- MCP client over stdio, SSE and streamable-HTTP with an approval step
Desktop & Deployment
Electron 33 · Docker · Kubernetes · Cloudflare Pages- Packaged builds for macOS, Windows and Linux with auto-update
- Docker Compose and Kubernetes manifests included
- Runs on 2 vCPU and 4 GB comfortably; 1 vCPU and 2 GB is the floor
Nothing exotic sits in this stack: Remix, Express, PostgreSQL and literal SQL statements, with no ORM and no proprietary framework for your team to master first. A new provider is two files; a new plan is a type union plus a config row, so your engineers can take ownership from the first day. The in-browser runtime is StackBlitz WebContainers, which StackBlitz licenses commercially, so any for-profit product using it in production must hold a StackBlitz license that the operator obtains.
Build to Scale
Built to Scale - Stateless Runtime, Pooled Reads & Space to Grow
Most of what feels expensive in a code-generation product - editing, file operations, preview, terminal - costs you nothing here, because it happens in the user’s browser. Your infrastructure carries only AI relay traffic and database access.
Stateless AI Runtime
workerd instances hold no session state and replicate freely behind a load balancer.
- The Express tier scales horizontally behind a load balancer
- Cloudflare Pages deployment and a stateless runtime make geographic distribution a deployment decision rather than a rewrite
Zero Preview Infrastructure
Browser-side execution means preview and sandbox cost scales with your users’ machines, not your cloud bill.
- That inverts the usual scaling problem
- PostgreSQL is the one component that needs real capacity planning, and it is handling six tables
Pooled Database Access
Connections capped at ten with idle and connect timeouts.
- Every lookup path - email, username, session token hash, transactions by user - carries an index
- There is no ORM and no proprietary framework layer
- Every database access is a literal SQL statement
Graceful Degradation
If PostgreSQL becomes unavailable, the auth proxy returns 503 for authentication endpoints while continuing to forward everything else.
- Users running on their own provider keys keep working through the outage
- Provider model catalogs cache for five minutes with a static fallback, so an upstream outage does not empty the model picker mid-session
- Each streaming attempt carries a 45-second timeout with up to two automatic retries and exponential backoff
- On unrecoverable failure the pipeline emits a structured error annotation rather than dropping the stream, so the client can surface something useful
Documented Benchmarks
The technical documentation states 500 concurrent users and 200 concurrent streaming sessions on a single instance, with p95 database query latency of 8ms and 2.3ms of auth-proxy overhead per request.
- Idle memory is 128MB for the proxy and 256MB for workerd, growing roughly 5MB per active streaming session
- Store hydration from localStorage completes in under 50ms
Built to Be Tailored and to Scale Over Time
The parts of this product most likely to change - which models you offer, what they cost, who can use them - are configuration, not code.
- Every provider is a class implementing one contract
- Every plan is a type union plus configuration rows
- Your engineers can read the whole data path in an afternoon
- The parts most likely to be audited are plain enough to actually audit
Honest scaling note: Rate limiting is currently an in-memory map, which means limits reset on restart and are not shared across instances. Moving to a Redis-backed limiter is documented as the multi-instance step, and we will tell you that before you scale out, not after.
07Security and trust
Built to Be Audited
Bolt.new Clone Security - OWASP-Mapped, Proxy-Isolated & Independently Assessed
The platform ships with a formal VAPT and compliance report mapped against the OWASP Top 10 (2021), NIST CSF 2.0 and SOC 2 control objectives – not a security page, an actual assessment with findings, severities and remediation status. The strengths it identifies as exceeding industry norms at this maturity level: parameterized queries throughout via postgres.js tagged templates, session tokens carrying 256 bits of entropy and stored only as SHA-256 hashes, timing-safe constant-time comparison for admin credentials, SSRF protection on the git proxy with domain whitelisting and private-IP and metadata-endpoint blocking, server-side credit validation that client tampering cannot reach, a separate admin session with its own cookie and validation path, JSON body parsing scoped to specific routes rather than applied globally, and rate limiting on authentication endpoints.
- 01 · SessionsProtected Sessions, Queries & Provider Keys
Sessions, database access and provider credentials are guarded at the server. Session tokens carry 256 bits of entropy and are stored only as SHA-256 hashes, and database credentials live solely in the Express process, where the workerd runtime cannot reach them.
- Parameterized postgres.js queries across all six tables, with no string-built SQL
- Operator API keys added at the proxy and absent from every browser response
- Passwords hashed with bcrypt at 12 rounds and inputs validated at the API boundary
- Parameterized Queries Throughout: Every database access uses postgres.js tagged templates. The assessment found no string-interpolated SQL across any of the six tables.
- Session Tokens Hashed at Rest: 256 bits of entropy from crypto.randomBytes, stored as a SHA-256 hash. Database read access does not yield a usable session.
- Keys Never Reach the Browser: Operator provider keys are injected server-side at the proxy and are never present in any response body sent to the client.
- 02 · AccessAdmin Authentication & Plan-Based Access
Users sit on Free, Pro or Enterprise plans, and the operator console runs on a separate admin session with its own cookie and a 24-hour expiry. Admin credentials are compared in constant time, and with no admin password configured, admin login is disabled rather than defaulting.
- 45+ feature flags re-validated server-side, so an edited cookie changes nothing
- Provider, model and template access gated per plan by the operator
- Timing-Safe Admin Comparison: A constant-time XOR comparison for admin credentials, with no default password - if none is configured, admin login is disabled entirely.
- SSO and two-factor authentication available as add-ons, not enabled out of the box
- 03 · AbuseRate Limits, Abuse & SSRF Protection
Authentication endpoints are rate limited, and outbound git requests pass through a proxy that only reaches approved hosts. The current limiter is in memory and resets on restart, so a Redis-backed limiter is the documented step before running multiple instances.
- SSRF guard with domain whitelisting and wildcard matching on repository imports
- Private IP ranges and cloud metadata endpoints blocked at the git proxy
- Credit balances checked on the server, out of reach of client tampering
- SSRF-Protected Git Proxy: Domain whitelisting with wildcard matching, private IP range blocking, and cloud metadata endpoint protection.
- 04 · IsolationExecution Isolation, Logging & Hardening
Generated code never executes on your infrastructure, MCP tool calls pass an approval step, and every credit movement is written to an append-only ledger. Five security headers ship on every response; CSP, HSTS and persistent audit logging are named hardening steps you add before going live.
- WebContainers keep npm installs, builds and previews inside the user's browser
- Admin credit adjustments recorded with a mandatory reason in the transaction ledger
- Five Security Headers Shipped: X-Frame-Options, X-Content-Type-Options, X-XSS-Protection, Referrer-Policy and Permissions-Policy on every response. CSP and HSTS are deployment additions.
- Disable demo accounts and add CSP, HSTS and encryption at rest before launch
- 05 · AssessmentIndependently Assessed Security Practices
The Miracuves Bolt.new Clone comes with a formal VAPT report that maps its controls to the OWASP Top 10 (2021), NIST CSF 2.0 and SOC 2 control objectives, plus a developer security handbook. It is an assessment for your reviewers to read, not a certificate.
- Each finding listed with its severity and current remediation status
- Strengths and open items both stated, so reviewers see the full picture
- Mapped to control objectives as evidence for your review, with no certification claimed
- 06 · PrivacyPrivacy Built for GDPR Obligations
Generated projects run in the user's browser, so your servers only relay AI requests and keep platform records such as accounts, plans and the credit ledger. That narrows what you hold, but GDPR duties such as lawful basis, data subject requests and retention policy stay with you as the operator.
- Builds, installs and previews happen client-side rather than on your infrastructure
- Accounts can be deactivated instantly while their history stays intact
- Retention, export and deletion automation is built on request, not shipped by default
What is configuration required before production: Content-Security-Policy and HSTS headers are not set by default and must be added. Demo accounts must be disabled. Two-factor authentication, database and provider-key encryption at rest, persistent audit logging, CSRF token validation and admin IP restriction are documented hardening steps rather than shipped defaults. We state this because it is in the assessment, and you would find it in a security review anyway.
Speak With Us on Security & ComplianceTrust and Safety
Bolt.new Clone Trust and Safety - What the Software Provides and What Stays in Your Hands
A Bolt.new clone is software, not a code reviewer, a moderation team or a legal opinion. Miracuves builds and hands over the governance layer an AI app builder runs on. What your users generate, how you police it and the law in each market you serve stay with you as the operator.
“Ready” means the controls are in the code: per-plan provider and model access, token ceilings and credit limits, 45+ server-validated feature flags, instant account suspension with history kept, an approval step on MCP tool calls, an SSRF-guarded git proxy, admin credit adjustments written to the ledger with a reason, and generated code that runs in the user’s browser rather than on your servers. It is not a certification. The VAPT report maps controls to OWASP, NIST CSF 2.0 and SOC 2 objectives for your own review; Miracuves holds no certificate on your behalf.
What the Product Supplies
Plan tier sets the providers, models and templates open to each user. Credit usage is written to an append-only ledger. The package also brings suspension of accounts, feature flags, a step where MCP use needs approval, and the separate admin session mentioned previously. It is all delivered as source code you own.
What Remains on the Operator
Your terms of service and acceptable-use policy for what users may build, the people who act on abuse, and your agreements with each AI provider whose keys you use. A StackBlitz commercial license for WebContainers is also yours to obtain, as the tech stack note says.
Generated Code Is Reviewed by People
The AI writes code; it does not vouch for it. Every change can be reviewed as a visual diff before it is accepted, but checking generated code for security, licensing and fitness before it ships is the operator’s or the end user’s responsibility, not the platform’s.
Jurisdiction Decides the Duties
Duties differ by market: GDPR in the EU and UK for the accounts and ledger you hold, the EU AI Act’s transparency duties where users interact with an AI system, and India’s Digital Personal Data Protection Act 2023. Confirm yours with counsel. Miracuves does not provide legal or regulatory advice.
08Grow the platform
- Prompt-to-app generation
- 22 LLM providers
- Credit metering
- Owner workspace
How add-ons work
- 1Choose the modules
Choose them on day one or bolt them on later; nothing underneath gets rebuilt.
- 2Each module scoped & priced
Every module gets its own scope and written quote.
- 3Shipped to your repo
This part too is delivered as owned source code, the same as the core.
Go Further
Bolt.new Clone Add-Ons - Mobile, Payments, Compliance & Enterprise Scale
Nothing is missing from the platform, and it can be shown exactly as it ships. Operators usually add the modules below for one of two reasons: a third-party account has to be set up, or a security review, procurement process or sheer scale asks for it.
Hover to hold it still · drag the strip to browse
- 01Product reach
Native Mobile Applications
There is no mobile app today. Native iOS and Android builds for project monitoring and review are available as custom development.
Mobile · Native - 02Billing
Payment Processing
The platform meters and prices usage but does not process payments. Stripe or another provider is wired in at deployment, including webhook signature verification.
Payments · Billing - 03Safeguards & hardening
SSO & Two-Factor
SAML or OIDC single sign-on and TOTP two-factor authentication are additive. Neither ships enabled, and the assessment lists 2FA as a planned control.
Auth · SSO - 04Safeguards & hardening
Audit Logging & Retention
A dedicated audit datastore for admin actions, plus retention, export and deletion automation, are recommended in the documentation and built on request.
Audit · Compliance
- 05Scale
Distributed Rate Limiting
Redis-backed rate limiting and cache for multi-instance deployments, replacing the in-memory map that resets on restart.
Scale · Distributed - 06Safeguards & hardening
CSP & HSTS Hardening
Content-Security-Policy and Strict-Transport-Security are not set by default. Both are named as immediate deployment actions and we configure them on request.
Headers · Hardening - 07Safeguards & hardening
Encryption at Rest
Database-level encryption and application-side encryption of stored provider keys are documented hardening steps rather than shipped defaults.
Encryption · At Rest - 08Product reach
Team Collaboration
Real-time collaborative editing, shared team workspaces and project version history are roadmap items available as custom scope.
Collaboration · Teams
More AI & No-Code Platforms
Launching an AI Product? Our Coverage Spans the Complete Market.
The Bolt.new Clone is one platform in a broader AI and no-code suite. If your roadmap extends beyond code generation, these connect naturally.
01Conversational AI platform with threaded chat, model routing, prompt libraries and usage-based billing.
Explore Solution
02Drag-and-drop website builder with a visual editor, templates, hosting and domain management.
Explore Solution
03Design-led site builder with commerce, blogging, scheduling and integrated hosting.
Explore Solution
04On-device AI reference resolution for context-aware assistants and conversational interfaces.
Explore Solution09The business case
Market Benchmarks
AI App Builder Market Benchmarks - Independent 2025-2026 Data
The revenue models listed above are ours, whereas the statistics that follow are not. They were published by JetBrains, Stack Overflow, GitHub, together with the Stanford Institute for Human-Centered AI, and are reproduced here as published. Each claim about how widely developers use AI to build software, how many AI projects exist and how much capital is flowing in can therefore be checked against sources you open yourself.
84% of Developers Use or Plan to Use AI Tools
Stack Overflow's 2025 Developer Survey, announced on 29 July 2025 from more than 49,000 responses in 177 countries, reports that 84 percent of developers use or plan to use AI tools in their development process. It also finds that 46 percent do not trust the accuracy of AI output, and that nearly 77 percent say vibe coding, generating software from prompts, is not part of their professional work.
4.3m AI-Related Repositories on GitHub in 2025
GitHub's Octoverse 2025 report, published on 28 October 2025, finds that AI-related repositories on GitHub now exceed 4.3 million, nearly doubling in less than two years. The same report says 80 percent of new developers on GitHub use Copilot in their first week, and that more than 36 million developers joined the platform in the past year.
85% of Developers Regularly Use AI Tools
JetBrains' State of Developer Ecosystem 2025, published on 15 October 2025 and based on 24,534 developers surveyed from April to June 2025, finds that 85 percent regularly use AI tools for coding and development, and 62 percent rely on at least one AI coding assistant, agent or code editor. Another 15 percent have not yet adopted AI tools in their daily work.
$581.7bn of Global Corporate AI Investment in 2025
The Stanford Institute for Human-Centered AI's 2026 AI Index reports that global corporate AI investment reached US$581.7 billion in 2025, up 130 percent from the prior year. The report also finds that generative AI reached 53 percent population adoption within three years, faster than the personal computer or the internet.
Stack Overflow, Stack Overflow's 2025 Developer Survey Reveals Trust in AI at an All Time Low, press release, 29 July 2025 - stackoverflow.co
GitHub, Octoverse: A new developer joins GitHub every second as AI leads TypeScript to #1, 28 October 2025 - github.blog
JetBrains Research, The State of Developer Ecosystem 2025, 15 October 2025 - blog.jetbrains.com
Stanford HAI, Inside the AI Index: 12 Takeaways from the 2026 Report, 13 April 2026 - hai.stanford.edu
Each figure is a publisher's own survey result, research-report statistic or reported total, repeated unchanged. They describe the market rather than how any platform built by Miracuves performs, and none of them projects revenue for your business.
The Financial Rationale
Marketability, Revenue Potential & Business Horizon
An AI code generation platform is not only a developer tool – it is a metered infrastructure business. Your cost is provider tokens; your revenue is credits, seats or licenses. The margin lives in the spread, and the admin console is what lets you tune it.
- 01
Metered by Design
Credit accounting per model per plan means your pricing can track provider economics instead of guessing at them.
- 02
Low Marginal Infrastructure
Preview and execution run on the user’s machine. Your per-user infrastructure cost is AI relay and a database row.
- 03
Provider Arbitrage
Route cheap workloads to fast open-source models and premium requests to frontier models, priced accordingly on every tier.
- 04
Procurement-Ready Governance
Central keys, per-plan access and feature flags are what enterprise buyers ask for before they ask about features.
- Pro and Enterprise plan subscriptions
- per-model input and output credit rates
- markup on the shared provider keys users spend
- seat licensing for Enterprise accounts
- white-label reselling to agencies and vendors
- bonus credit blocks for trials and retention
Operated with discipline, the Miracuves Bolt.new Clone can turn into:
- 01a metered AI infrastructure business
- 02a branded prototyping engine for agencies
- 03an internal developer platform that keeps code in-house
- 04an app builder embedded in your SaaS
These are business patterns, not promises. Actual outcomes depend on your audience, provider costs, credit pricing and go-to-market. What the platform gives you is the control surface to run any of these models.
Paying users on Pro plans, with usage metered in credits.
Income blends plan fees with the gap between what providers charge you and what you charge per credit. Free users who bring their own keys add very little cost.
Studios run it under their own name for client delivery.
Enterprise seat licensing with unlimited credits fits buyers who value quick prototypes and centrally held keys more than per-token billing.
A platform team rolls AI-assisted development out to every engineer.
This one is never offered to outside customers. Value shows up as engineering time saved and code that stays inside your network, which is often what gets the budget approved.
10Why Miracuves
Why Miracuves
Miracuves vs Other Bolt.new Clone Developers
There are many ways to get an AI code generation platform: open-source forks, freelancers, agencies, or owning a documented product.
Why Founders Turn to Miracuves
- 01
The Commercial Layer Ships With It
Auth, subscription tiers, credit metering with a transaction ledger, and an operator console. Forking an open-source builder leaves every one of those as your problem.
- 02
Governance Is Real Screens
Provider keys, per-plan model access, credit pricing, user administration and 45+ server-validated feature flags - configurable at runtime, not in a config file.
- 03
Multi-Provider Done Properly
One BaseProvider contract across 22 SDKs with a documented key-resolution chain. Adding a provider is two files, not a refactor.
- 04
Total Source Ownership
The complete Remix, Express, workerd and PostgreSQL codebase is yours to modify and deploy. No per-seat fee, no vendor who can change terms.
- 05
Documented at Enterprise Level
PRD, feature catalog, ERD, schema interpretation, API collection, technical dossier, security handbook and a formal VAPT report - enough to survive procurement, not just a demo.
- 06
Plain Statements on Readiness
Included, Supported and Configuration required are stated per capability - including that CSP, HSTS, 2FA, payments and encryption at rest are deployment work. Nothing surprises you after purchase.
See How We Compare and Why Clients Rank Us the #1 Partner for Ready-Made Clone Solutions
| Criteria | MiracuvesMiracuves Bolt.new Clone | Generic Clone Script | Custom Dev Agency |
|---|---|---|---|
| Time to Launch | 6 days to deploy | Unknown / DIY | 6-9+ months |
| Source-Code Ownership | ✔ Full | Often limited / encrypted | Usually yes |
| Feature Depth (Bolt.new-like) | High (generation, provider routing & credit governance) | Basic (prompt-to-code only) | Depends on budget |
| Security & Compliance | VAPT report & security handbook | Minimal | Varies widely |
| Scalability & Performance | Cloud & CDN-optimized | Rarely considered | Depends on architecture |
| Monetization Options | Plans, per-model credits, seats, licensing | Limited / needs custom work | Custom (more time & cost) |
| Admin & Analytics | Full-fledged dashboard | Very basic or missing | Custom build (extra cost) |
| Cost vs Speed vs Quality | Balanced | Cheap but risky | High cost, slow |
| Ongoing Support & Updates | Available with clear plans | Usually none | Depends on contract |
Forking an open-source AI builder gets you a demo. What it does not get you is accounts, plans, credit metering, provider governance or anything a buyer will sign off on. Here is where we differ.
Talk to Our TeamChangelog
Bolt.new Clone Release Log - Version History & Updates
| Version | Date | What’s New |
|---|---|---|
| v2026.3 | Aug 2026 | Multiple bug fixes across generation, provider routing, credit deduction and the operator console. |
| v2026.2 | Jun 2026 | BYOK support added - users can supply their own provider API keys on any plan, alongside operator-managed shared keys. |
| v2026.1 | Mar 2026 | Initial release - AI code generation platform with 22 providers, browser runtime, credit metering, six deploy targets and an operator console. |
Let’s build together
Take your idea from concept to a platform that runs.
Schedule a free consultation to get a straightforward timeline and honest answers. We would rather earn your trust than rush you into a purchase.
- Reply in 24 hrs
- Free consultation
- No lock-in contracts
- Full source code ownership
11Learn more
Blog & Resources
Bolt.new Clone App - Latest Blog & Guides
Stay updated with the latest trends, guides and case studies on AI code generation, multi-provider LLM architecture and developer platform economics.





Product5 questions
01What is a Bolt.new Clone App?
A Bolt.new Clone App is a ready-made AI code generation platform modeled on the prompt-to-app experience Bolt.new made popular: users describe an application in natural language and receive a complete, runnable multi-file project that boots and previews inside their browser. The Miracuves Bolt.new Clone adds what a commercial deployment needs around that core - accounts and sessions, three subscription tiers, credit metering priced per model, 22 LLM providers behind one selector, six deployment targets and an operator console for providers, pricing, plans and feature flags - launched under your own brand.
02Does it include a mobile app?
No. The Miracuves Bolt.new Clone is browser-first, with packaged Electron desktop builds for macOS, Windows and Linux. There is no Android or iOS application. Native mobile is available from Miracuves as custom development if your roadmap needs it, and we would rather say so up front than let you find out later.
03Where does the generated code actually run?
In the user's browser. The Miracuves Bolt.new Clone runs projects in WebContainers, a complete Node.js environment client-side, so npm installs, shell commands, builds and previews all execute locally. Your servers only relay AI requests and store platform data such as accounts, plans and the credit ledger. For teams with IP or compliance constraints, this is usually the deciding factor.
04Can I add my own or private models?
Yes. Use the Ollama provider for locally hosted models, configure any OpenAI-compatible endpoint through provider settings, or expose additional models through the OpenRouter integration (365+ models). Adding an entirely new provider to the Miracuves Bolt.new Clone is two files - a class extending the base provider contract, and an entry in the registry.
05Do we receive the Bolt.new Clone code?
Yes. Miracuves hands over the complete Remix and React web application, the Express authentication proxy, the Cloudflare workerd runtime, the PostgreSQL schema across six tables, the provider abstraction layer covering all 22 providers, the Electron desktop packaging and every integration service, with full ownership to modify and extend. There is no revenue share or per-seat fee.
Pricing & go-live4 questions
06How much does it cost to build an AI app builder in 2026?
The Miracuves Bolt.new Clone starts at $3,699 for a launch-ready white-label deployment. Scope beyond that depends on your provider strategy, how you design credit pricing, the integrations you add and your infrastructure. One cost sits outside our price: the in-browser runtime is StackBlitz WebContainers, which StackBlitz licenses commercially, so a for-profit product using it in production needs a StackBlitz license that the operator obtains directly.
07When can the Bolt.new Clone launch?
Our side takes six working days. The Bolt.new Clone is already built, so the work comes down to applying your branding and deploying to your server. Time extends only while we wait for you to supply brand assets, hosting access, provider API keys, or choices on plan structure and credit pricing.
08How does the credit system work?
Each model has input and output credit rates you configure per plan. After a generation completes, the platform computes the cost from the actual token counts, validates the balance server-side, deducts it and writes an append-only ledger entry recording the amount, the resulting balance, the model and a description. Enterprise accounts bypass metering entirely.
09Does it handle payments and subscriptions?
It handles the subscription model - tiers, entitlements, credit allocations and the full transaction ledger. It does not process payments. There is no built-in checkout, so connecting Stripe or another payment provider is operator work at deployment, and Miracuves can wire it in, including webhook signature verification.
Protection & oversight4 questions
10How are provider API keys protected?
Operator keys live in the database and are injected into requests by the Express proxy at the server layer. They are never included in any response sent to the browser, and users only ever see keys they supplied themselves. Encryption of those keys at rest is a documented hardening step rather than a default, and Miracuves flags it before purchase rather than after.
11Can the Bolt.new Clone pass our security due diligence?
The Miracuves Bolt.new Clone ships with a formal VAPT and compliance report mapping its controls to the OWASP Top 10 (2021), NIST CSF 2.0 and SOC 2 control objectives, alongside a developer security handbook. It lists real strengths - parameterized queries throughout, SHA-256 hashed session tokens with 256 bits of entropy, timing-safe admin comparison, SSRF protection on the git proxy, server-side credit validation and separate admin sessions - and is equally explicit about what remains configuration required. It is an assessment for your reviewers, not a certificate.
12What do I need to configure before going live?
A Content-Security-Policy header and HSTS, both deployment additions rather than defaults. Demo accounts disabled. Your own provider API keys and a strong admin password. If you are billing, a payment provider with webhook verification. At scale you will also want Redis-backed rate limiting, database and key encryption at rest, and persistent audit logging. Every one of these is named in the documentation as configuration required, and Miracuves configures them on request.
13Is it legal to launch a Bolt.new clone app?
Making software that works like Bolt.new is a common practice across the AI app-builder category. Our own team built the Miracuves Bolt.new Clone, and nothing in it is copied from Bolt.new sites or apps, be it code, design, graphics or content. It launches under your brand, with your pricing, your choice of models and your policies. The Bolt.new name is a trademark held by its owner. Miracuves gives no legal advice, so run your launch plan past your own lawyer.
Why Miracuves1 question
14Why choose Miracuves to build a Bolt.new clone?
Miracuves has completed 9,000+ projects across 6,000+ clients and lists 90+ ready-made clone solutions, among them a no-code and AI suite that sits beside this Bolt.new Clone. Where a fork stops short, this product adds the commercial layer: auth, tiers, an operator console and credit metering backed by a ledger. A VAPT report and full documentation come with it, and go-live takes 6 working days. Support includes 60 days of platform guidance, priority bug fixes for 6 months, and updates for 12 months. Each company figure, along with the scope it covers, sits on the Miracuves facts page.
Go deeper on this platform
Jump to an industry hub, a development service, long-form guides and the one-time cost.
SaaS development
See the service→Talk to an expert →Disclaimer
Miracuves is an independent software development company. We are not affiliated with, connected to, sponsored by, or endorsed by Bolt.new.
Read our full Legal Notice & Disclaimer →“Bolt.new Clone” is used descriptively. It is how the software industry refers to building a platform with functionality similar to Bolt.new, and how clients search for it.
The entire design and codebase is built by our own team. The product contains no code, design, graphics, or content originating from the Bolt.new website or applications.
Bolt.new and all other third-party names and marks are the property of their respective owners, referenced here solely to describe the category of software offered.





















