# CatDoes Cloud
Source: https://docs.catdoes.com/catdoes-cloud
The backend infrastructure that powers your CatDoes apps — automatically configured through AI.
## Overview
CatDoes Cloud is the managed backend that runs behind every app you build. You don't need to set up servers, databases, or authentication manually — CatDoes configures everything through your natural language prompts.
## What's included
A managed database for your app's data. CatDoes creates tables, relationships, and queries based on your app's requirements.
User sign-up, login, and session management — configured automatically. Supports email/password, social logins, and more.
File and media storage for user uploads, images, and app assets. Managed and scaled for you.
Serverless functions for custom backend logic, webhooks, and third-party integrations — deployed at the edge for low latency.
## How it works
When you describe your app, CatDoes automatically provisions the backend resources you need. For example:
* "I need users to sign up and log in" → authentication is configured
* "Users can upload profile pictures" → storage bucket is created
* "Show a feed of posts" → database table and API endpoints are set up
You can also make changes later by describing what you need in conversation, and CatDoes Cloud updates accordingly.
CatDoes Cloud is fully managed. You don't need to worry about scaling, backups, or infrastructure maintenance.
# Base44 to Native App
Source: https://docs.catdoes.com/features/base44
Convert your Base44 website into a native iOS and Android app.
Base44 generates standard web applications using React and Tailwind. To get them onto mobile, most wrapping tools place your site inside a **WebView** — an embedded browser that makes the web app *look* native but still runs through a browser engine. This adds latency, blocks access to hardware features like Bluetooth, NFC, Face ID, and real push notifications, and often gets rejected from the App Store for not offering functionality beyond a website. A true native app talks directly to the operating system's UI components with no browser layer in between, which is what CatDoes builds for you.
There are two ways to convert your Base44 website into a native app. Pick whichever works best for you.
## Method 1: Using Your Website URL
Best for public-facing websites where all pages are visible without logging in.
1. Visit [catdoes.com](https://catdoes.com) and enter your Base44 website URL.
2. Log in to CatDoes — the system automatically detects your website.
3. Choose whether to make your project **Public** (free) or **Private** (requires subscription).
4. CatDoes analyzes your website, extracting:
* Design elements and color scheme
* Branding (logo, fonts, icons)
* Content from your pages
* Screenshots for reference
5. The AI agent creates your mobile app specification based on your website's design.
6. The software agent builds your native iOS and Android app.
Your website is analyzed automatically. You don't need to provide any design files or assets — everything is extracted from your live site.
## Method 2: Import via GitHub
Best when your Base44 website has pages behind a login — like dashboards, account pages, or admin panels — that the agent can't see by visiting the URL. By importing through GitHub, the agent gets full access to your entire project and can rebuild every part of your app accurately.
### Step 1: Connect Base44 to GitHub
1. If you don't already have a GitHub account, go to [github.com](https://github.com) and sign up — it's free and only takes a minute.
2. In your Base44 project, go to the **GitHub** section and connect your project to a repository. Base44 will push your project's code to GitHub automatically.
### Step 2: Import into CatDoes
1. Go to [app.catdoes.com](https://app.catdoes.com/dashboard) and click **Create New App**.
2. Choose **Import from GitHub**.
3. Authorize CatDoes to access your GitHub account and select the repository that contains your Base44 project.
4. Pick the branch you want to import (usually the default one is fine).
5. Give your project a name and click **Import**.
CatDoes will pull in your entire project and set everything up. Once the import is done, the AI agent has full access to all your code — including pages that were behind a login — and can start building your native mobile app from there.
## After Conversion
Once your app is created, you can:
* Preview it in your browser or with [Expo Go](/features/live-preview)
* Continue refining with the AI agents
* Deploy to [web, Android, or iOS](/features/deployment)
The conversion process captures your website's current state. After conversion, you can ask the AI agents to customize the app further or add mobile-specific features like push notifications, offline support, or native gestures.
# Checkpoints
Source: https://docs.catdoes.com/features/checkpoints
Every new chat saves a snapshot of your project so you can roll back anytime.
## What Are Checkpoints?
Every time you start a new chat, CatDoes automatically saves a checkpoint — a snapshot of your project at that moment. If something goes wrong in a later chat, you can roll back to any previous checkpoint and pick up from there.
Think of it like save points in a game. Each new chat is a new save. If things don't work out, you can always go back.
## How It Works
1. You're working on your project in a chat.
2. You start a new chat — CatDoes saves a checkpoint of your project's current state.
3. You keep building in the new chat.
4. If something breaks or you don't like the direction, you roll back to the checkpoint from before that chat started.
You don't need to do anything extra — checkpoints are created automatically whenever you start a new chat.
## When to Start a New Chat
Since each new chat creates a checkpoint, starting a new chat is also how you create a save point. Some good moments to do that:
* **After finishing a feature you're happy with.** Start a new chat so that state is saved.
* **Before trying something risky.** About to experiment with a big layout change or a new feature? Start a new chat first so you can roll back if it doesn't work out.
* **When the conversation gets long.** Long chats can cause the agent to lose track of earlier context. Starting fresh gives you a clean slate and a save point.
## Rolling Back
To roll back to a previous checkpoint:
1. Click on **History** in the chat panel.
2. Browse and select the chat you want to roll back to.
3. Click **Revert to this point** to restore your project to that checkpoint.
Your project will return to the exact state it was in when that chat started. This means you can experiment freely — try bold changes, test ideas, let the agent go in a new direction — knowing you can always undo it all in one step.
## Tips
* **Start a new chat when you're happy.** If your project looks good right now, start a new chat to lock that in as a checkpoint.
* **Don't be afraid to experiment.** Checkpoints mean you can always undo. Try things out without worrying about breaking what you've built.
* **Roll back instead of fixing.** If the agent went in the wrong direction and the result is far from what you want, rolling back and re-prompting is often faster than trying to fix it piece by piece.
# Custom Domain
Source: https://docs.catdoes.com/features/custom-domain
Use your own domain name for your deployed web app.
When you deploy your project, CatDoes gives you a free URL like `yourapp.catdoes.app`. If you want to use your own domain instead (like `www.yourcompany.com`), you can set that up too.
## How It Works
### Step 1: Deploy Your Project
First, ask the AI agent to deploy your project. Just tell the agent something like "I want to deploy my project" and it will walk you through picking a URL and getting your app live at `yourapp.catdoes.app`.
If you haven't deployed yet, check out the [Deployment](/features/deployment) page for more details.
### Step 2: Ask for a Custom Domain
Once your project is deployed, tell the agent you want to use your own domain. For example, you can say "I want to use my custom domain" and the agent will guide you through the setup.
### Step 3: Set Up Your Domain
You'll need to go to wherever you manage your domain (like GoDaddy, Namecheap, Google Domains, Cloudflare, or any other provider) and point it to your CatDoes app. Here's what to do:
1. Log in to your domain provider.
2. Find the **DNS settings** for your domain.
3. Add a new **CNAME** record:
* **Name/Host**: your domain or subdomain (for example `www` or `app`)
* **Value/Target**: `yourapp.catdoes.app` (replace `yourapp` with the slug you chose during deployment)
4. Save the changes.
DNS changes can take a few minutes to a few hours to take effect, depending on your domain provider.
### Step 4: Come Back and Confirm
Once you've set up the DNS record, go back to CatDoes and let the agent know. The agent will verify the setup and get your custom domain connected.
## Tips
If you're not sure where your DNS settings are, search for "DNS settings" in your domain provider's help pages — every provider has them, they're just in different places.
You can always keep using your `yourapp.catdoes.app` URL even after setting up a custom domain. Both will work.
# Debug Mode
Source: https://docs.catdoes.com/features/debug-mode
How to prompt the agent to find and fix bugs in your project.
## When Something Breaks
When you hit a bug, tell the agent to enter debug mode. The agent will add console logs throughout the relevant code, and then you reproduce the broken flow on your device. The logs give the agent real output to inspect — making it much easier to pinpoint what's broken and where.
Here's the prompt:
> "Enter debug mode: add logs wherever necessary, I will redo the broken flow on my device so the logs appear for you to inspect. Giving more insight into what's broken and where the bug is."
## The Flow
1. **You prompt debug mode.** Tell the agent what's broken and ask it to add logs.
2. **The agent adds logs.** It reads through the relevant code and places console logs in the areas that matter.
3. **You reproduce the bug.** Redo the broken flow on your device so those logs fire.
4. **The agent inspects and fixes.** With real log data, the agent can see exactly where things went wrong and apply the fix.
You don't need to know where the bug is — you just need to trigger it. The logs do the talking.
## Give Context
The more you tell the agent about the problem, the better it knows where to place logs. Include:
* **What's broken** — what you see happening (or not happening).
* **What you expected** — what should have happened instead.
* **Any extra context** — when it started, what changed recently, or any error messages you see.
For example:
> "Enter debug mode. The contact form stopped submitting after we added the new header. Add logs to the form submission flow — I'll try submitting again and share what comes back."
> "Debug mode. The checkout page crashes when I click pay. Here's the error I'm seeing: \[paste error]. Add logs around the payment flow so I can reproduce and get more detail."
## Use Screenshots with Markers
If the bug is visual, take a screenshot and mark the problem area with a red circle or arrow. Upload it with your debug prompt:
> "The section I circled in red is overlapping the footer. Enter debug mode and add logs — I'll reload the page and share the output."
A marked screenshot removes all ambiguity — the agent can see exactly what you're looking at. Use any basic markup tool (your phone, Preview on Mac, Paint on Windows, or any screenshot tool) to draw on the screenshot before uploading.
## Share Error Messages
If you see an error message — in the preview, in a popup, or in your browser's developer console — paste it directly into your prompt. Error messages help the agent know exactly where to place logs and often point straight to the root cause.
> "I'm getting this error when I submit the form: `TypeError: Cannot read property 'email' of undefined`. Enter debug mode, add logs around this, and I'll reproduce it."
## Tips
* **Say "debug mode."** This signals the agent to investigate thoroughly — read the code, add logs, and prepare for you to reproduce the issue.
* **Mention what changed.** If something broke after a recent edit, say so. It helps the agent place logs in the right spots.
* **One bug at a time.** If you have multiple issues, debug them in separate prompts. This keeps the logs focused and leads to cleaner fixes.
* **Always share the log output.** The whole point of debug mode is the back-and-forth — the agent adds logs, you reproduce, and the logs reveal the problem.
* **Don't worry about being technical.** You don't need to understand the code. Just describe what's wrong, reproduce the flow, and paste what you see.
# Deployment
Source: https://docs.catdoes.com/features/deployment
Deploy your CatDoes app to the web, Android, and iOS.
When your app is ready, just tell the agent where you want to release it — for example, *"I want to release my app to the App Store"* or *"Deploy my app to the web."* The agent takes care of the rest, walking you through each step right in the chat.
## Web
Deploy your app to a `*.catdoes.app` URL and share it with anyone.
1. Tell the agent you want to release your app for the web.
2. Pick a short URL slug for your app (e.g., `myapp` becomes `myapp.catdoes.app`).
3. CatDoes builds and deploys your app — you'll get a live link when it's done.
You can change your slug later. The old URL is automatically cleaned up.
**Slug rules:** 1-30 characters, lowercase letters, numbers, and hyphens only.
Free-plan web deployments include a small "Powered by CatDoes" badge.
## Android
Build your app for Android and publish it on Google Play.
1. Tell the agent you want to release your app for Android.
2. A secure input field appears for your Expo token — the agent guides you through creating one if you don't have it yet.
3. Confirm your app version and choose a build type.
4. CatDoes builds your app — this usually takes 10 to 30 minutes.
5. Once the build is ready, you can download it and publish it on Google Play yourself.
Android releases require a paid plan and an [Expo](https://expo.dev) account.
## iOS
Build your app for iOS and submit it to the App Store.
1. Tell the agent you want to release your app for iOS.
2. Secure input fields appear for your Expo token and your Apple Developer credentials — the agent guides you through setting these up if it's your first time.
3. Confirm your app version and CatDoes starts the build — this usually takes 10 to 30 minutes.
4. When the build is done, you can ask the agent to submit it directly to the App Store.
5. From there, you decide on Apple's side whether to release it through **TestFlight** for testing or make it available publicly.
iOS releases require a paid plan, an [Expo](https://expo.dev) account, and an [Apple Developer](https://developer.apple.com) account.
Your credentials are entered through secure input fields that encrypt everything you type. The AI never sees your actual tokens or passwords — it only knows whether they've been provided or not.
# Free Credits
Source: https://docs.catdoes.com/features/free-credits
Ways to earn free CatDoes credits — from sharing on X to community programs.
## Share on X
You can earn up to **500 free credits** by sharing your CatDoes project on X (Twitter). Look for the free credits option in your dashboard to get started.
## CatDoes Catapult
CatDoes Catapult is a monthly program where we hand-pick users and give them **free credits and hands-on support**. Every month, we select builders based on what they're working on — whether it's a side project, a startup, or something experimental.
If you're selected, you'll receive free credits to keep building plus direct support from the CatDoes team to help you move faster.
**How to apply:**
1. Join the [CatDoes Discord](https://discord.com/invite/g9zaWq5wby).
2. Introduce yourself in the **#introduction** channel — tell us who you are and what you're building.
3. Fill out the [application form](https://forms.gle/Fosvjp4otA7VxspZ6).
We review applications each month and reach out to selected users directly.
## Cat Composer
Cat Composer is a special role we give to active members of the CatDoes Discord community. If you're regularly helping others, sharing feedback, or contributing to discussions, you may be selected to become a Cat Composer.
**What you get as a Cat Composer:**
* **Free monthly credits** to use on your projects
* **Early access to experimental features** before they're available to everyone
* **Direct access to the founders** — ask questions, share ideas, and get support straight from the team
Join the [CatDoes Discord](https://discord.com/invite/g9zaWq5wby) and get involved to be considered.
# GitHub Integration
Source: https://docs.catdoes.com/features/github-integration
Connect your GitHub repository so AI agents can read your code, make changes, and push updates directly.
Connect your CatDoes project to a GitHub repository. Once connected, AI agents can read your existing code, make changes, commit them, and push updates to your repository — all while maintaining full git history.
GitHub integration requires an active paid subscription.
## Connecting Your Repository
1. From your project dashboard, click **Connect GitHub**.
2. On GitHub, authorize the CatDoes app and select which repositories to grant access to.
3. Back on CatDoes, select the repository you want to link from the dropdown.
4. Choose the branch you want to work with (defaults to your repository's default branch).
5. Click **Link Repository**.
Your GitHub integration is now active. AI agents can now read, modify, and push code to the selected branch.
## Working with Branches
You can switch branches or create new ones directly from your project dashboard.
**To switch branches:**
1. Click **Change** next to your connected repository
2. Select a different branch
3. Click **Update**
**To create a new branch:**
1. In the branch dropdown, click **Create New Branch**
2. Enter a branch name (like `feature/new-feature` or `fix/bug-name`)
3. Optionally select which branch to create it from
4. Click **Create**
The new branch is created on GitHub and immediately available for use.
## How AI Uses GitHub
When you ask an AI agent to make changes:
1. The agent reads your existing code from the connected repository
2. Makes the requested modifications
3. Commits the changes with a descriptive message
4. Pushes to your selected branch
All changes are tracked in your git history, just like regular commits.
The agent can also:
* Pull the latest changes from GitHub before making modifications
* Handle merge conflicts if they arise
* Reset to the remote state if needed
## Changing or Disconnecting
**To change repository or branch:**
1. Go to your project dashboard
2. Click **Change** on the GitHub Integration card
3. Select a new repository and/or branch
4. Click **Update**
**To disconnect:**
1. Go to your project dashboard
2. Click **Disconnect** on the GitHub Integration card
Disconnecting removes the link but doesn't affect your repository on GitHub.
## Tips
Work on feature branches instead of main. This keeps your main branch stable while letting the AI experiment on separate branches.
Review AI commits on GitHub after they're made. Check the diff to see exactly what changed and why.
# Import from GitHub
Source: https://docs.catdoes.com/features/import-from-github
Bring an existing GitHub project into CatDoes and let AI turn it into a mobile app.
Already have a project on GitHub? You can import it directly into CatDoes instead of starting from scratch. This is a great way to bring in projects from tools like Lovable, Bolt, or any codebase you already have on GitHub.
Importing from GitHub requires a Plus plan or above.
## How to Import
1. Go to [app.catdoes.com](https://app.catdoes.com/dashboard) and click **New Project**.
2. In the dialog, click **Import from GitHub**.
3. You'll be taken to GitHub to authorize CatDoes. Choose which repositories you want to give CatDoes access to, then confirm.
4. Once you're redirected back to CatDoes, the import form opens automatically.
5. Select your **repository** from the dropdown.
6. Choose the **branch** you want to import (your default branch is selected automatically).
7. Give your project a **name** — it's auto-filled from your repository name, but you can change it.
8. Choose your project visibility:
* **Public** — anyone can see your project. Free to create.
* **Private** — only you and your team can see it. Requires an active subscription.
9. Click **Import**.
CatDoes will pull in your code and set everything up. Once it's done, you're taken straight to your new project, ready to start working with the AI agents.
## When to Use This
* You built something with **Lovable**, **Bolt**, or another tool and want to turn it into a native mobile app
* You already have a codebase on GitHub and want AI agents to work with it
* Your project has pages behind a login that CatDoes can't access by visiting the URL directly — importing the code gives the agent full access to everything
## Tips
If your project isn't on GitHub yet, most tools (like Lovable) let you connect to GitHub with a few clicks. Once your code is pushed there, you can import it into CatDoes.
After importing, your project is linked to the GitHub repository. You can manage the connection from your project dashboard — see [GitHub Integration](/features/github-integration) for details.
# Live Preview
Source: https://docs.catdoes.com/features/live-preview
Preview your app in real time as you build it with CatDoes.
## Browser Preview
As you build your app with CatDoes, a live browser preview appears in a panel alongside your conversation. Every change the agent makes — new screens, design tweaks, feature additions — is reflected in the preview automatically.
### Auto-Refresh
The preview refreshes automatically when:
* The agent finishes writing or modifying code files
* A codebase scan completes
* The agent explicitly reloads the preview after installing packages or making structural changes
You can also manually refresh the preview at any time using the **refresh button** in the preview panel.
### Soft Reload
A soft reload refreshes the preview without restarting anything. It's a quick way to reconnect and see recent changes.
**When to use it:**
Sometimes the agent disconnects from the preview and you can't see the latest changes. A soft reload reconnects it and brings everything up to date.
**How to soft reload:**
* **In the browser preview** — click the refresh button in the preview panel.
* **In Expo Go** — hold three fingers on the screen to open the reload menu.
* **Ask the agent** — tell the design or software agent to soft reload. For example:
> "Soft reload the preview"
### Hard Reload
A hard reload fully restarts the development server. It clears cached data and gives your project a fresh start.
**When to use it:**
If something isn't working and you can't figure out why — the page looks stuck, a change isn't showing up, or the app is behaving strangely — a hard reload often fixes it. These are usually cache-related problems that go away with a restart.
**How to hard reload:**
Ask the agent to hard reload for you. For example:
> "Hard reload"
If you've been trying to fix an issue and nothing seems to work, try a hard reload before spending more time debugging. It clears out cached data and restarts fresh — many mysterious problems turn out to be a simple cache issue.
### Runtime Errors
If your app encounters a runtime error during preview, CatDoes captures it and makes it available to the agent. The agent can inspect these errors and attempt to fix them automatically as part of its workflow. For more on how the agent monitors and responds to errors, see [CatDoes Watch](/features/watch).
## Mobile Preview with Expo Go
Beyond the browser preview, you can test your app on a physical mobile device using [Expo Go](https://expo.dev/go).
1. Download **Expo Go** on your iOS or Android device.
2. Use the **Scan QR Code** option in the preview panel to get a QR code for your project.
3. Scan the QR code with your device to open the app in Expo Go.
This lets you test touch interactions, native features, and real-world performance directly on your phone.
The browser preview renders your app in a web view, which may look or behave slightly differently from the native mobile experience. Use the Expo Go preview on a physical device to verify native-specific behavior.
Use the QR code to share your in-progress app with teammates or stakeholders for quick feedback — no install required, just the Expo Go app.
# Lovable to Native App
Source: https://docs.catdoes.com/features/lovable
Convert your Lovable website into a native iOS and Android app.
CatDoes can transform your existing Lovable website into a native mobile application. The process extracts your website's design, branding, and content to create a fully native iOS and Android app.
There are two ways to convert your Lovable website into a native app. Pick whichever works best for you.
## Method 1: Using Your Website URL
Best for public-facing websites where all pages are visible without logging in.
1. Visit [catdoes.com/lovable](https://catdoes.com/lovable) and enter your Lovable website URL.
2. Log in to CatDoes — the system automatically detects your website.
3. Choose whether to make your project **Public** (free) or **Private** (requires subscription).
4. CatDoes analyzes your website, extracting:
* Design elements and color scheme
* Branding (logo, fonts, icons)
* Content from your pages
* Screenshots for reference
5. The AI agent creates your mobile app specification based on your website's design.
6. The software agent builds your native iOS and Android app.
Your website is analyzed automatically. You don't need to provide any design files or assets — everything is extracted from your live site.
## Method 2: Import via GitHub
Best when your Lovable website has pages behind a login — like dashboards, account pages, or admin panels — that the agent can't see by visiting the URL. By importing through GitHub, the agent gets full access to your entire project and can rebuild every part of your app accurately.
### Step 1: Connect Lovable to GitHub
1. If you don't already have a GitHub account, go to [github.com](https://github.com) and sign up — it's free and only takes a minute.
2. In your Lovable project, go to the **GitHub** section and connect your project to a repository. Lovable will push your project's code to GitHub automatically.
### Step 2: Import into CatDoes
1. Go to [app.catdoes.com](https://app.catdoes.com/dashboard) and click **Create New App**.
2. Choose **Import from GitHub**.
3. Authorize CatDoes to access your GitHub account and select the repository that contains your Lovable project.
4. Pick the branch you want to import (usually the default one is fine).
5. Give your project a name and click **Import**.
CatDoes will pull in your entire project and set everything up. Once the import is done, the AI agent has full access to all your code — including pages that were behind a login — and can start building your native mobile app from there.
## After Conversion
Once your app is created, you can:
* Preview it in your browser or with [Expo Go](/features/live-preview)
* Continue refining with the AI agents
* Deploy to [web, Android, or iOS](/features/deployment)
The conversion process captures your website's current state. After conversion, you can ask the AI agents to customize the app further or add mobile-specific features like push notifications, offline support, or native gestures.
# Plan Mode
Source: https://docs.catdoes.com/features/plan-mode
How to prompt the agent to plan your project before it starts building.
## Prompt the Agent to Plan First
The agent can build things fast, but it builds what you ask for — so if your first prompt is "build me a dashboard," it'll start building immediately. A better approach is to prompt the agent to **plan first**, so you can align on what gets built before any code is written.
This matters most when you're:
* Starting a new project from scratch
* Adding a major feature with multiple parts
* Working on something you haven't fully figured out yet
## How to Plan with the Agent
Tell the agent what you want to build and explicitly ask it to plan before building. The key phrase is: **"Don't build anything yet."**
Here are some examples:
* "I want to add a booking system to my site. Plan out the full feature — what pages I need, how the user flow should work, and what components to build. Write it all down before you start coding."
* "I'm building a dashboard with user analytics. Break down the structure, list the sections, and outline what data each section shows. Don't build anything yet."
* "Plan a checkout flow for my store. Think through each step the user goes through, what happens on each page, and how errors should be handled."
The more context you give — your goals, preferences, pages you already have, features you want — the better the plan will be.
## Use a Higher Tier for Planning
The **High** and **Max** agent tiers are best for planning. They're the most capable agents, so they're ideal for thinking through complex ideas, asking the right questions, and writing a detailed plan.
Once the plan is ready, you can switch to **Low** or **Medium** to actually build it. This way you get top-quality thinking upfront without spending top-tier credits on the entire build.
## Walk Through the Plan Together
Don't just read the plan and say "go." Treat it like a conversation:
1. **Start broad.** Tell the agent the big picture — "I want an online store" or "I need a portfolio site with a blog."
2. **Let the agent break it down.** It will outline the pages, components, and flow.
3. **Correct what doesn't match.** If the agent suggests a three-step checkout but you want a single-page checkout, say so now.
4. **Agree on the order.** Decide what gets built first. Usually it makes sense to go page by page or feature by feature.
Changing a plan is free. Changing built work costs credits.
## Tips
* **Say "write the plan down."** Be explicit about this. When the plan is written in the conversation, the agent can reference it while building — and so can you.
* **Don't worry about technical terms.** Describe things the way you'd explain them to a friend. The agent will translate your ideas into a buildable plan.
* **Review before building.** Read through the plan carefully. If something doesn't match what you had in mind, correct it now — it's much easier to fix a plan than to redo built work.
* **Plan in pieces for big projects.** If your project has multiple major parts, plan and build one part at a time. Planning everything at once can lead to a plan that's too long for the agent to follow precisely.
# Saving Credits
Source: https://docs.catdoes.com/features/saving-credits
Tips and strategies to get the most out of your CatDoes credits.
## How Credits Work
Every time a CatDoes agent performs a step — generating a design, writing code, or running a build — it consumes credits. The cost of each step depends on which **agent tier** you're using and the complexity of the task.
## Pick the Right Agent Tier
CatDoes lets you switch between **Low**, **Medium**, **High**, and **Max** agent tiers. This is the single biggest lever you have for controlling credit usage. Higher tiers think harder and produce higher-quality results, but they cost more credits per step.
| Tier | Best for | Relative cost |
| ---------- | --------------------------------------------------------------------- | ------------- |
| **Low** | Simple tasks, small visual changes, repetitive work | Lowest |
| **Medium** | Bigger features, detailed designs, fixing tricky problems | Moderate |
| **High** | Best-quality thinking, planning complex features, important decisions | High |
| **Max** | The hardest problems where top quality matters most | Highest |
New projects start on the **Medium** tier, a balanced default that handles most work well.
The **Max** tier delivers the deepest reasoning and best-in-class quality, but it costs the most per step. A great way to use the higher tiers without burning through credits is to **let them plan first**. Ask **High** or **Max** to figure out how a feature should be built — the approach, the structure, the steps — and then switch to **Low** or **Medium** to actually build it. You get top-tier thinking upfront, and the bulk of the work happens at a lower cost. This often leads to better results in fewer steps, saving you credits overall.
Start with **Low** or **Medium** and only move up to **High** or **Max** when the task genuinely requires it. For many routine tasks — small layout adjustments, simple component creation, or basic CRUD logic — the lower tiers deliver solid results at a fraction of the cost.
The **High** and **Max** tiers require a credit balance above 50. **Max** is available only when your balance is above 200. If your balance drops below these thresholds, top up your credits to unlock them again.
You can switch tiers from the agent dropdown in your project dashboard. The change applies to the next step the agent takes.
## Write Clear Prompts
Vague instructions lead to more back-and-forth steps, and each step costs credits. The more specific you are upfront, the fewer steps the agent needs to get it right. See the [Writing Good Prompts](/features/writing-good-prompts) page for detailed tips and examples.
## Start a New Chat When the Conversation Gets Long
The longer the conversation, the less sharp the agent becomes — it might forget things, make mistakes, or break something that was working. Starting a new chat gives the agent a clean slate so it works better and wastes fewer credits.
Click the **Compact and Save** button at the bottom of your conversation. Your project and codebase carry over. If something goes wrong, you can always **roll back** to a previous chat to restore your codebase and conversation history.
## Earn Free Credits
There are several ways to earn free credits — from sharing on X to community programs like CatDoes Catapult and Cat Composer. See the [Free Credits](/features/free-credits) page for details.
## Quick Reference
* Use **Low** agents for simple tasks, **Medium** for complex ones, and **High** or **Max** only when top-tier quality is essential.
* Write specific, detailed Prompts to reduce unnecessary steps.
* Start a **new chat** when the conversation gets too long.
- **Share on X** to earn free credits.
# Tech Stack
Source: https://docs.catdoes.com/features/tech-stack
The technology behind apps built with CatDoes.
## React Native Expo
Every app built with CatDoes uses [React Native Expo](https://expo.dev/) under the hood. This gives you:
* **Cross-platform** — One codebase for iOS, Android, and web.
* **Native performance** — Your app uses native UI components, not a web wrapper.
* **Expo ecosystem** — Access to a rich set of libraries for camera, location, notifications, and more.
## What this means for you
* If you ever want to eject and continue development on your own, you have a standard React Native Expo project.
* Your app isn't locked into CatDoes — you own the code.
* Engineers on your team can extend the app using familiar React Native tooling.
You don't need to know React Native to use CatDoes. The AI handles all the code. But if you do, you can always dive into the source.
# CatDoes Watch
Source: https://docs.catdoes.com/features/watch
Built-in error monitoring for your CatDoes apps across web, iOS, and Android.
## Overview
CatDoes Watch is an integrated error monitoring system that captures, groups, and manages runtime errors across your web, iOS, and Android apps. It is built into the CatDoes dashboard — no third-party tools or separate SDKs to configure.
## Features
* **Multi-platform error capture** — Collects runtime errors from web, iOS, and Android apps with full stack traces, file locations, and device information.
* **Intelligent issue grouping** — Errors are automatically grouped into issues using server-side fingerprinting, so duplicate occurrences of the same error appear as a single issue with an event count.
* **Issue management** — Resolve, reopen, ignore, or delete issues directly from the dashboard.
* **Environment filtering** — View errors from development, production, or both environments separately.
* **Privacy-safe by default** — Sensitive data like emails, passwords, API keys, and credit card numbers are automatically redacted before storage.
## Dashboard
Navigate to the **Watch** tab in your project dashboard to see an overview of all captured issues. The dashboard shows:
* **Stats bar** — Unresolved issues, total issues, events in the last 24 hours, and total events.
* **Issues table** — Each row shows the error title, source location, event count, platform, environment, and when it was last seen.
* **Issue detail panel** — Click any issue to open a side panel with the full stack trace, individual event history, and device metadata.
### Filtering and sorting
You can filter issues by status and environment, and sort by different criteria:
* **Status** — Unresolved (default), All, Resolved, or Ignored.
* **Environment** — All, Development, or Production.
* **Sort by** — Last seen, Most events, or First seen.
## Settings
Open **Watch > Settings** from the dashboard to configure error capture. The settings page includes:
### API key
Your API key authenticates your app with the Watch ingestion endpoint. Keys are shown only once after generation — copy it immediately. You can regenerate a key at any time if it is compromised.
### Capture toggles
Control which errors are sent to Watch:
| Toggle | Description |
| ------------------ | ----------------------------------------------- |
| Development errors | Capture errors from local previews and staging |
| Production errors | Capture errors from published production builds |
| Console errors | Forward `console.error` calls to Watch |
### Data retention
Set how long error data is retained, up to the maximum allowed by your plan.
## Plan limits
Watch usage limits depend on your CatDoes plan:
| Plan | Events/min | Events/day | Retention |
| --------------- | ---------- | ---------- | --------- |
| Free | 10 | 100 | 1 day |
| Basic / Starter | 30 | 500 | 7 days |
| Pro | 50 | 500 | 30 days |
| Business | 100 | 2,000 | 90 days |
If you hit your rate limit, new events are dropped until the limit resets. Upgrade your plan for higher limits and longer retention.
# Writing Good Prompts
Source: https://docs.catdoes.com/features/writing-good-prompts
How to write clear, effective prompts that save credits and get better results from CatDoes agents.
Every message you send to a CatDoes agent kicks off a series of steps — designing layouts, building features, and refining your app. The clearer your prompt, the fewer steps the agent needs to get it right, and the fewer credits you spend.
## Why Prompt Quality Matters
CatDoes agents work step by step. Each step consumes credits. A vague prompt forces the agent to guess what you mean, try different approaches, or ask follow-up questions — all of which use up credits without moving your project forward.
A well-written prompt does the opposite: it gives the agent a clear target so it can get things right on the first try.
## Be Specific About What You Want
The most common mistake is being too broad. Compare these two prompts:
**Vague:**
> "Make the homepage look better."
**Specific:**
> "Redesign the top of the homepage with a large centered title, a short description below it, and a 'Get Started' button. Use the brand colors we already have."
The vague prompt leaves the agent guessing — which part of the homepage? What does "better" mean? The specific prompt tells it exactly what to build and how it should look.
## Describe What You Want to See
Focus on the end result — what should it look like when it's done? You don't need to know anything technical. Just describe what you'd want to see on the screen, and the agent will handle the rest.
**Instead of:**
> "Fix the layout."
**Try:**
> "I want three pricing cards sitting next to each other, evenly spaced, centered on the page."
Think of it like describing a room to an interior designer. You say "I want a cozy reading corner with a lamp and a bookshelf" — you don't need to tell them which screws to use.
## Upload a Mood Board or UI Inspiration
If you have screenshots, mood boards, or examples of designs you like, upload them and tell the agent what they are. This is one of the most effective ways to get great design results.
For example:
* Collect a few screenshots of websites or apps whose style you like
* Upload them and say: "This is my mood board. I want my site to have a similar feel — clean layout, lots of white space, modern typography."
* Or upload a single screenshot and say: "I like how this site does their hero section. Design mine with a similar layout but using my brand colors."
You can also upload rough sketches or wireframes you've drawn — even on paper. The agent will understand the layout you're going for and build from there.
The more visual context you give, the less the agent has to guess. A mood board with three or four reference images plus a sentence about what you like about each one gives the agent a clear direction.
## Describe the Look and Feel
The biggest gap between what you want and what the agent designs comes down to your prompt. You have a picture in your head — the agent doesn't. The more you can describe that picture, the closer the result will be.
Instead of only saying where things go, tell the agent what vibe you're going for. Words like "modern," "playful," "corporate," "minimal," or "bold" go a long way.
**Example:**
> "I want a landing page that feels clean and professional — something like a fintech company. Mostly white space with a bold headline."
If you have brand colors or preferences, state them upfront. Otherwise the agent will pick something reasonable, but it might not match your vision.
* Colors: "Use dark blue and white as the main colors."
* Layout: "Two columns on a computer screen, but stacked on top of each other on a phone."
* Style: "Keep it clean and simple — no flashy effects."
* Reference: "Something like the Stripe homepage — modern and minimal."
You don't need design terminology. Everyday descriptions like "clean," "bold," "playful," or "professional" work well.
## Break Large Tasks Into Smaller Ones
If you need a full page with a navigation bar, a welcome section, feature highlights, testimonials, and a footer — don't ask for it all at once. Break it into focused requests:
1. "Add a navigation bar at the top with our logo on the left and menu links on the right."
2. "Below the navigation bar, add a welcome section with a big heading, a short tagline, and a signup button."
3. "Below that, show three features side by side, each with an icon and a short description."
Smaller, focused prompts keep the agent on track and reduce the chance of it misunderstanding what you want. You can review and adjust each part before moving on.
## Mention What's Already There
When you're changing something that already exists in your project, tell the agent what to keep. This helps it avoid accidentally removing or replacing things you like.
**Good:**
> "Add a phone number field to the contact form, right below the email field. Keep everything else the same."
**Less helpful:**
> "Add a phone number field."
The agent looks at your project before making changes, but giving it context in your prompt helps it make better decisions.
## Iterate Instead of Starting Over
If the first result is close but not quite right, tell the agent what to change rather than starting from scratch.
**Instead of:**
> "That's not what I wanted. Redo it."
**Try:**
> "I like the layout, but make the heading bigger and change the background to light gray."
Small adjustments cost fewer credits and build on work that's already done.
## Quick Reference
| Do | Don't |
| -------------------------------------------------- | ------------------------------------------------- |
| Describe what you want to see on the screen | Give vague or one-word instructions |
| Upload mood boards or reference images | Expect the agent to read your mind |
| Break big requests into smaller ones | Ask for an entire page in one message |
| Mention what's already there that you want to keep | Assume the agent remembers everything from before |
| Describe colors, layout, and style you prefer | Leave the entire look up to the agent |
| Iterate on what's close | Start over when small tweaks would fix it |
# Quick Start
Source: https://docs.catdoes.com/getting-started
Go from an idea to a running mobile app in minutes with CatDoes.
## Build your first app
What used to take months of hiring designers, developers, and going back and forth — you can now do in minutes. Just describe your idea, and CatDoes takes care of the rest.
Head to [catdoes.com](https://catdoes.com) and create a free account. No credit card required.
Start a new project and tell CatDoes what you want to build. You don't need a full spec — even a single sentence is a great starting point. For example: *"a multiplayer trivia game where friends compete in real-time"* or *"a booking app for my barbershop where clients can pick a barber, choose a time, and pay upfront."* The AI will ask follow-up questions to fill in the gaps.
As your app takes shape, you can see it live in your browser or scan a QR code to try it on your phone. Want to change the color scheme? Add a new screen? Just say so in the chat and watch it update.
When you're happy with your app, publish it directly to the App Store, Google Play, or the web — all from CatDoes. No extra tools or setup needed.
Start simple. Focus on the core idea of your app first, then add more features as you go. You can always iterate.
# Welcome to CatDoes
Source: https://docs.catdoes.com/index
Build mobile apps with AI — just describe what you want.
## What is CatDoes?
CatDoes is an AI-powered platform that turns your app ideas into real, production-ready mobile applications. No coding required — just describe what you want to build, and CatDoes handles the design, development, and deployment for you, all through a simple conversation.
Describe your app idea in plain language. CatDoes asks the right questions, designs the screens, writes the code, and wires up your backend — all in one conversation.
See your app come together in real time. Preview it in the browser or scan a QR code to try it on your phone as it's being built.
CatDoes asks for your input on key decisions and checks with you before making big changes. It's your app — the AI just builds it faster.
Publish directly to the App Store, Google Play, and the web. No extra tools, no manual setup.
## Get started
Go from idea to a running app in minutes.
Learn about the backend infrastructure that powers your apps.
## Platform highlights
See your app in a live browser preview with a QR code to test on your phone.
Release directly to iOS, Android, and the web — no extra tooling needed.
Built-in error monitoring and crash reporting with AI-powered fix suggestions.
Your apps are built with React Native Expo — a proven, battle-tested framework.
## Who is CatDoes for?
Whether you're a founder with a startup idea, a designer who wants to ship without a dev team, or an engineer looking to move faster — CatDoes meets you where you are.
Start building for free. Upgrade when you're ready.