Privacy Policy

Last updated: October 24, 2026

1. Who we are

Capacity is operated by Rocket Social, a French SASU (société par actions simplifiée unipersonnelle) with a share capital of €500, registered office 21 rue Valentin Couturier, 69004 Lyon, France, registered with the Lyon Trade and Companies Register under number 832 329 445 (SIRET 832 329 445 00023, VAT FR88832329445) ("Capacity", "we", "us"). This policy covers the Capacity website and web app at capacity.so, Capacity Desktop, the Capacity mobile app, and Capacity App Builder, the Capacity plugin you can connect to AI assistants such as ChatGPT and Claude (together, the "Service"). Rocket Social is the data controller for the personal data described here.

Rocket Social also publishes other products, including Breakreach (see rocketsocial.org). In the ChatGPT plugin directory, Capacity App Builder is listed under Rocket Social's verified developer account, which is named Breakreach.

Questions or requests about your data: support@capacity.so.

2. Capacity App Builder at a glance

Capacity App Builder lets ChatGPT or another AI assistant create, edit, preview and publish the web apps of your Capacity account when you ask it to. In short:

  • What we collect: the inputs of the tools the assistant calls, which are limited to your apps (app names, file paths, the app's code, commands run in the app's workspace, version IDs), plus the connection between the assistant and your Capacity account. Section 4 lists every tool's inputs and outputs.
  • What we never collect: your conversations with the assistant, your assistant account's profile or memory, your location, payment card details, health information, government ID numbers, biometric data, or passwords and API keys.
  • Why: to build, preview, host and publish your apps, to keep the Service secure and working, and to measure how the tools are used.
  • Who receives it: the assistant's provider (for example OpenAI for ChatGPT), which receives the tool results, and the hosting, logging and analytics providers listed in section 4. Capacity does not send your tool inputs to any AI model, and does not sell them or use them for advertising.
  • How long: your apps' code and versions as long as your account exists; call logs 30 days; error reports 90 days; usage statistics 25 months; the connection until you disconnect.
  • Your controls: disconnect at any time in Settings → Connectors on capacity.so (or in the assistant's own settings), restore or unpublish an app, and delete your account and its data in Settings → Delete account.

3. Data we collect

Account data

  • Name, email address and password. Passwords are stored only as a salted hash (bcrypt), never in readable form.
  • If you sign in with Google or Apple: the name, email address, profile picture and language that the provider shares with us.
  • Optional profile and onboarding details you choose to give: company, company size, goal, how you heard about us, city, country, language and social profile links.
  • Your plan, subscription status and, if you were referred, who referred you.

Payment data

Payments are processed by Stripe and Polar on their own checkout pages. We never receive or store your card number. We receive a customer identifier, your plan, the amount paid and your billing country. When you start a checkout on capacity.so, we pass your email address, IP address, browser user agent and ad-click identifiers to the payment provider for fraud prevention and to measure our advertising. Capacity App Builder never handles payments.

Content you create

  • Your prompts and conversations with the Capacity AI agent in the Capacity editor, including images you attach.
  • The code, files and version history of your apps, and the environment variables you set for them.
  • Files you upload, and files uploaded by your apps.
  • The data stored in your published apps' databases (see section 11).

Connected services

When you connect a service such as Supabase, Apple App Store Connect or Google Play, we store the access tokens or keys needed to act on your behalf, encrypted. When you connect an AI assistant, see section 4.

Usage and device data

When you use our website and app we collect pages visited, clicks and other interactions, session recordings (with everything typed into form fields masked), IP address, browser, device type and the page that referred you, through the analytics tools listed in section 8. Our servers and hosting providers also record technical logs (such as the time and IP address of requests) and error reports to keep the Service secure and working.

Support and feedback

Messages you send us by chat or email, feedback, and answers to the optional survey shown when you cancel a subscription.

4. Capacity App Builder in ChatGPT, Claude and other AI assistants

This section describes exactly what happens with your data when you connect Capacity App Builder to an AI assistant. The assistant's provider (for example OpenAI for ChatGPT or Anthropic for Claude) processes your conversation, and the tool results we send back, under its own privacy policy.

Connecting

The assistant connects through OAuth: you sign in to your Capacity account (with your email and password, or Google or Apple, on capacity.so, never in the chat) and approve the connection. From then on, the assistant sends an access token with each request so that its tools act only on your account and your apps. Using Capacity App Builder requires a Capacity account with a plan that includes it; it never asks for payment details.

What each tool receives and returns

The assistant decides when to call a tool, based on your request. These are all the tools, with everything they receive from the assistant and send back to it:

list_apps

Receives: An optional filter on app names.

Returns: Your apps' IDs, names and types, whether they are published, and their preview, production and editor URLs.

create_app

Receives: The new app's name and type (static or full-stack).

Returns: The new app's ID and URLs, and the stack guide (the coding rules of the app's template).

get_app

Receives: The app's ID.

Returns: The app's URLs and status, the stack guide, the list of its files and its five most recent versions.

read_files

Receives: The app's ID, up to 20 file paths and an optional range of lines.

Returns: The content of those files of the app.

search_code

Receives: The app's ID, a search pattern and an optional filter on file paths.

Returns: The matching lines, with their file name and line number.

apply_changes

Receives: The app's ID, a short summary of the change, and the files to create, edit or delete with their new content (the app's code and text).

Returns: Whether the app built or restarted, the build errors, type errors or crash logs to fix, the files changed and the new version's ID.

run_command

Receives: The app's ID, a command (for example to install a code library) and an optional time limit.

Returns: The command's exit code and output. Commands run only inside the app's own cloud workspace.

get_logs

Receives: The app's ID and how many lines to return.

Returns: The app's recent server logs and its last build log. These logs contain whatever your app prints, which can include data your app handles.

screenshot

Receives: The app's ID, which page of the app to capture and whether to capture the full page.

Returns: An image of that page of the app's preview, and its URL.

list_versions

Receives: The app's ID.

Returns: Up to 30 recent versions of the app: ID, summary and date.

restore_version

Receives: The app's ID and the version to go back to.

Returns: Confirmation that the app was restored.

publish

Receives: The app's ID and an optional site title and description, which become public (search engines and link previews show them).

Returns: The app's public production URL.

set_env_vars (not available in ChatGPT)

Receives: The names and values of the environment variables you ask the assistant to set for an app.

Returns: The names that were set. Values are never returned.

We do not receive or request your conversation history, your assistant account's profile, memory or files, your location, or any other information from the assistant.

No sensitive or restricted data

Capacity App Builder never asks for, and none of its tools has a field for, payment card details, health information, government ID numbers (such as social security numbers), biometric data, or passwords, API keys and other secrets. In ChatGPT, when an app needs a secret, you add it yourself in the Capacity editor under Environment Variables, never in the chat. The assistant is also instructed not to build apps that collect payment card numbers, medical records, government ID numbers or biometric data: for payments, an app links to a checkout page hosted by the payment provider.

What we keep, why, who receives it and for how long

Connection: the assistant's name (for example "ChatGPT"), its redirect addresses, hashed access and refresh tokens, the Capacity account they belong to and when they were last used.

Purpose: Recognise the assistant and act only on your account.

Recipients: Our database host, MongoDB Atlas.

Retention: Until you disconnect. Access tokens expire after 30 days and are renewed until then; authorization codes expire after 10 minutes.

Your apps' code and versions: the files the assistant writes, each change's summary and the assistant's name.

Purpose: Build, preview, host and publish your apps, and let you restore any version.

Recipients: Google Cloud (cloud workspaces), GitHub (each app's private code repository), Cloudflare (previews and published sites), Fly.io and Turso (published full-stack apps and their databases).

Retention: As long as the app and your account exist; deleted within 30 days of your account's deletion. A workspace's disk is deleted 5 days after it stops (it stops after 45 minutes without activity); the code stays in the app's repository.

The site title and description sent with publish.

Purpose: Describe your published app to search engines and link previews.

Recipients: Public, on your published app.

Retention: Until you change them or unpublish the app.

Server logs of each call: the tool's name, your account ID, the assistant's name, the time and duration, and, when a call fails, the first 1,000 characters of the error (which can include build output). Tool inputs are not written to these logs.

Purpose: Keep the Service working and fix errors.

Recipients: Google Cloud (logging).

Retention: Up to 30 days.

Error reports when a call fails unexpectedly, which can include the request that failed.

Purpose: Find and fix bugs.

Recipients: Sentry.

Retention: Up to 90 days.

Usage statistics: which tool was called, by which assistant, simple counts (such as the number of files changed or the app's type) and, when your plan doesn't cover the request, your plan's name. Linked to your account ID.

Purpose: Understand how Capacity App Builder is used and improve it.

Recipients: PostHog (EU region).

Retention: Up to 25 months.

Screenshots.

Purpose: Let the assistant check what your app looks like.

Recipients: ScreenshotOne, which receives the URL of the preview page to capture.

Retention: Capacity does not keep them: the image goes straight back to the assistant.

The tool results we send to the assistant (file contents, logs, command output, screenshots, URLs) go to its provider, which keeps them with your conversation under its own policy. Capacity does not send tool inputs or outputs to any AI model: the code is written by your assistant.

Your controls

  • Disconnect: in Settings → Connectors on capacity.so, which deletes the assistant's tokens immediately, or from the assistant's own settings.
  • Undo a change: restore any earlier version of an app, from the assistant or in the Capacity editor.
  • Take an app offline: unpublish it in the Capacity editor.
  • Get your code: download it from the Capacity editor.
  • Delete everything: Settings → Delete account on capacity.so deletes your account, your apps (their code, databases and files) and your connections, as described in section 9. To delete a single app or other specific data, write to support@capacity.so.

5. Sensitive data

Capacity does not need, and does not ask for, sensitive information such as health data, biometric data, government identifiers (for example social security numbers) or payment card numbers. Please do not include such information in your prompts, in requests to an AI assistant, or in your app's code. Card payments for your Capacity subscription happen only on Stripe's or Polar's checkout pages.

6. How we use your data

  • To provide the Service (performance of our contract with you): creating, building, hosting and publishing your apps, running the Capacity AI agent and the tools your assistant calls, and keeping your version history.
  • To manage your account and billing (contract and legal obligations): authentication, subscriptions, invoices and accounting records.
  • To keep the Service secure and reliable (legitimate interest): preventing fraud and abuse, including on published apps, debugging errors and monitoring the Capacity AI agent.
  • To communicate with you (contract and legitimate interest): service emails, support, and product news you can unsubscribe from at any time.
  • To understand and improve Capacity (legitimate interest, or consent where the law requires it): product analytics and measuring our own advertising campaigns. Data from Capacity App Builder is not used for advertising.

We do not sell your personal data, and we do not use it to train AI models.

7. Automated processing

We do not make decisions with legal or similarly significant effects about you based solely on automated processing.

8. Who receives your data

We share personal data only with the service providers we need to run Capacity, each limited to its role:

  • Hosting and infrastructure: Google Cloud, Vercel, Cloudflare, Fly.io, Scaleway, Turso, MongoDB Atlas (our database host), GitHub (private repositories that store your apps' code) and Expo (mobile app builds).
  • AI model providers: Anthropic, OpenAI, Google, Mistral AI and OpenRouter to generate code for the Capacity AI agent, and Replicate and FAL to generate images. They receive the prompts, code and context needed for each request. This does not apply to Capacity App Builder, whose code is written by your own assistant.
  • AI agent monitoring: Flowlines, which receives traces of Capacity AI agent sessions (prompts and outputs) so we can debug them.
  • Payments: Stripe and Polar.
  • Email: Brevo (your name, email, company, location, language, plan and subscription status, to send service and marketing emails) and Google (some account emails).
  • Analytics and advertising measurement: PostHog (EU region), Google Analytics, Hotjar and Meta (Meta Pixel, and the Conversions API, which receives your email address in hashed form when you subscribe).
  • Error monitoring: Sentry.
  • Support chat: Crisp.
  • Screenshots: ScreenshotOne, which receives the preview URL of the app page being captured.
  • AI assistants you connect, which receive the tool results described in section 4.
  • Authorities and successors: when the law requires it, or to a successor in case of a merger or acquisition.

Apps you publish are public at their URL, and files you upload to an app are served at unguessable public links.

Some of these providers process data outside the European Union, in particular in the United States, where our cloud workspaces and servers run. In that case the transfer relies on appropriate safeguards such as the European Commission's Standard Contractual Clauses or the EU-US Data Privacy Framework.

9. How long we keep it

  • Account data, apps, code, conversations and files: as long as your account exists, then deleted within 30 days of your account's deletion.
  • Cloud workspaces: stopped after 45 minutes without activity; the workspace disk is deleted 5 days after it stops (your code stays in the app's repository).
  • Published apps: online until you unpublish them, your subscription ends or you delete your account.
  • AI assistant connections: until you disconnect them, which deletes their tokens immediately. Authorization codes expire after 10 minutes.
  • Server logs: up to 30 days.
  • Error reports: up to 90 days.
  • Analytics data: up to 25 months.
  • Invoices and accounting records: 10 years, as required by French law.
  • Support conversations: up to 3 years after your last contact.

10. Your choices and rights

  • Correct: update your profile in your account settings.
  • Export: download your app's code from the editor, or ask us for a copy of your personal data.
  • Delete: Settings → Delete account on capacity.so, or in the Capacity mobile app, or ask us at support@capacity.so. We then delete your account, apps and connections as described in section 9, except the invoices we must keep.
  • Disconnect AI assistants and API keys: Settings → Connectors.
  • Stop marketing emails: use the unsubscribe link in any of them.
  • Cookies: block or delete them in your browser settings (see section 12).

If you are in the European Economic Area, the United Kingdom or another region with similar laws, you also have the right to access, rectify, erase, restrict or object to the processing of your data, to data portability, and to withdraw your consent at any time. Write to support@capacity.so and we will answer within one month. You can also lodge a complaint with your data protection authority; in France, the CNIL (cnil.fr).

11. Data in the apps you build

When your published app collects data from its own users (for example in its database or through file uploads), we store and process that data on your behalf, as your processor. You are responsible for that data and for your app's own privacy policy.

12. Cookies

  • Essential: access_token keeps you signed in (30 days), and sidebar_state remembers your sidebar layout (7 days).
  • Analytics and advertising: cookies and local storage set by PostHog, Google Analytics, Hotjar and Meta, as described in section 8.
  • Support: cookies set by Crisp for the chat window.

Capacity App Builder sets no cookies in your assistant.

13. Security

Data is transmitted over HTTPS. Passwords and AI assistant or API key tokens are stored only as hashes, connection tokens to third-party services are encrypted, and access to production systems is restricted. No system is perfectly secure; if a breach affects your data, we will notify you and the competent authority as the law requires.

14. Children

Capacity is not directed to children under 13, or under the minimum age required in your country, and we do not knowingly collect their personal data. If you believe a child has given us personal data, contact us and we will delete it.

15. Google API Services User Data Policy

Capacity's use and transfer of information received from Google APIs to any other app will adhere to the Google API Services User Data Policy, including the Limited Use requirements. We only use this data to enable YouTube video uploads and do not transfer it to any third parties.

16. Changes to this policy

We may update this policy. We will post the new version on this page with a new date and, for significant changes, notify you by email or in the app.

17. Contact

Rocket Social SASU (Capacity), 21 rue Valentin Couturier, 69004 Lyon, France: support@capacity.so.