Patchy Cloud Privacy Notice
Last updated: November 10, 2026
This notice covers Patchy Cloud: the service at cloud.patchyhq.com, its sign-in and waitlist pages at accounts.patchyhq.com, and the patchy command line tool. The patchyhq.com website and PatchPage each have their own notice.
The short version: we store what your company publishes and saves, and reach the services it connects, such as its email or CRM, so that we can run its patches. We record who did what, so that we can operate, secure and improve the service. Patchy Cloud’s databases, file storage and server logs are in the United States. We do not sell personal information or share it for cross-context behavioral advertising, and we do not train models on your data.
Two kinds of information
- Your company’s content. Patches, tables, files, connections, your company’s look and what its patches read from the services it connects. We handle them on your company’s behalf. Your company decides what goes in and who sees it. Questions about that content go to your company’s admins first.
- Account and usage records. These are ours to look after, and the rest of this notice is mostly about them.
What we collect
- Account and company details. Your name, email address and how you sign in, held with Clerk, our sign-in provider, which also sends sign-in codes and invitations and sees the device and network details that signing in involves. When you accepted the terms and this notice. If you joined the waitlist, or used a link we sent you to skip it, your email address, and when you were invited and first signed in. Your company’s name and handle, its members, their roles and invitations, and its settings, such as whether agent access is on.
- Company content. What your company publishes and saves: its patches and their versions, tables, files and descriptions, its company look, and its connection settings, with their credentials encrypted.
- Data from the services your company connects. What its patches, and the agents of people building them, read from those services, such as database rows, analytics data, CRM records, and email with its attachments. We also keep a snapshot of each connection’s structure and settings, such as a database’s table names or an email address’s labels.
- Sign-in and setup records. The machines each person has logged in and the AI agents each person has connected, with details such as their names, when they were made and last used, and hashes of their tokens. Logging in from the command line tool suggests your computer’s name for the machine. The records Home uses to help you connect your agent.
- Activity and logs. Who published or changed each patch, and when. Who shared which patch with a guest at another company, when, and when the share ended. Which of your company’s patches you open, so that your Home can put your usual ones first and Patchy Cloud can have them ready to open. A log of the calls patches and agents make, which names the person and any agent, and can include query text and the subject and recipients of email a patch sends. The lines a patch writes to its own log, counts of the email each connection sends, short-lived records that stop an agent’s retry from repeating a change, and our servers’ own logs.
- Usage records. Records of requests and of key events, such as a person joining, a patch published or someone opening a colleague’s patch for the first time, that show how Patchy Cloud is used. They identify people by account id, not by name or email address. They can include the names a builder gave a patch’s parts, such as its tables, and details of the
patchycommand line tool and AI agents involved, such as the tool’s version, the coding agent it detected and an agent’s name. - Email you send us. Your address and whatever you write.
People who open a public patch
A public patch opens without an account. We count the visit and record the request like any other. If you are signed in to Patchy Cloud, the record carries your account id. If you are not, it carries no person’s id. Every patch runs in a sandbox designed to stop it loading anything from other sites or contacting them.
How we use it
- To run Patchy Cloud and show each person what they are allowed to see.
- To keep it secure, find faults and help when you ask.
- To understand how it is used, set its limits and, later, to bill for it.
- To understand what people build with it. For that we read usage records and each patch’s name and description.
- To tell you about changes to the service and these terms.
- To comply with the law and enforce the terms.
Who else handles it
We do not sell personal information or share it for cross-context behavioral advertising. We disclose it to the vendors below, each acting for us, and as your company instructs, as this notice describes.
- Amazon Web Services, US East: the servers that run Patchy Cloud and patch code, and their logs.
- Neon, US East: the databases and file storage.
- Clerk, US: accounts, sign-in, invitations and the waitlist.
- PostHog, US: usage records.
- Google: email you send us. Google may store it outside the United States.
- Anthropic and OpenAI: the AI tools our team uses to operate and repair the service. They can process company content only when we open it for a reason the terms allow, and our accounts are set so that what they process is not used to train models. They may process it outside the United States.
AI agents you connect. If your company turns agent access on and you connect an AI agent, Patchy Cloud gives it what it asks for through a patch’s agent tools, as you. That can include your company’s content and the names of people in it. What happens to it next is up to the agent. It may run on your computer or on its maker’s servers, and it may send what it reads to its maker. If it does, the maker’s terms, and the settings on your or your company’s account with them, decide how they keep and use it, including whether they train models on it. The same goes for what an agent reads while it builds a patch. Here the maker acts for you or your company, not as our vendor, even when it is one of the vendors listed above.
Companies that share a patch with you. If someone at another company shares a patch with you as a guest, that company’s members can see your name, email address and company, and which of its patches you have. Its patches can show your name and email address while the share lasts, and its logs record what you do in them. Your name stays in its records after the share ends, and your email address and company can still appear when its members share a patch again. Sharing a patch with an email address tells the person sharing whether the address has an active Patchy account.
Services your company connects. When an admin connects a database, a PostHog project, an Attio workspace, an email address on Pidgeon or a mailbox at your company’s email provider, we exchange credentials, requests and the data needed to carry out your company’s instructions with that service, and send email through Pidgeon if the admin allows. AI agents your people connect may also receive that data. Those services hold your company’s data under its own accounts and agreements with them. They work for your company, not as our vendors, even when, like PostHog, one is also listed above.
We may also disclose information where the law requires it or where it is needed to enforce the terms.
How long we keep it
We keep information while we need it to run and secure Patchy Cloud, and where the law requires it or to resolve a dispute or enforce the terms. How long depends on the kind of record:
- Company content: while your company keeps it. What it deletes is removed after a recovery or clean-up period.
- Call logs and connection snapshots: while your company exists on Patchy Cloud.
- Your account, machines and agent connections: until your account is removed, when you ask us or your company closes. Expired agent tokens can go sooner.
- Records of accepting the terms and this notice, such as who accepted and when: as legal records, while your account and your company exist and for up to 7 years after either closes, so that we can show what was agreed.
- Setup records: until your account is removed. Expired setup lines can go sooner.
- Which patches you open: until the patch or your account is removed for good.
- Sharing records: until the patch is removed for good, including after a share ends.
- Records that stop an agent repeating a change: about a day.
- Usage records: at PostHog, for as long as PostHog keeps them, so that we can see how use changes over time.
- Database history and server logs: until they cycle out. What is deleted elsewhere can remain in them until then.
When your company leaves, the terms say what we delete, and when.
Security
Connections to Patchy Cloud are encrypted, and so are the connection credentials it stores. Patch code runs in a sandbox designed to keep it apart from your sign-in and from other companies. No service is perfectly secure, and this one is a beta. If a security incident affects your information, we give the notices the law requires.
Your choices and rights
For anything inside a patch, ask your company’s admins. For your account, email us to ask what we hold about you, or to correct or delete it. Depending on where you live you may have further rights under laws such as the California Consumer Privacy Act. Email us and we will honour any that apply, and we will not treat you differently for asking.
Where it is processed
The beta is offered to companies in the United States, and Patchy Cloud’s databases, file storage and server logs are there. Copies that reach the vendors listed above are held where each vendor says. Google, Anthropic and OpenAI may process or store what reaches them in other countries. What your own AI agents read goes wherever those agents, and any makers they send it to, process it. Your company’s connected services keep its data wherever those services do, and email your patches send goes out through your company’s Pidgeon account.
Cookies and Do Not Track
Patchy Cloud uses cookies to sign you in and keep you signed in. Clerk sets those, including the ones it uses to protect sign-in from abuse. The load balancer in front of our servers, run by Amazon Web Services, also sets two cookies that keep your browser on the same server, so that a running patch stays connected. They expire seven days after your last visit. Patchy Cloud sets one of its own after you make a change such as inviting someone, so that the next page can say what was done. It holds your account id and that one line, which can name a person or a patch, and it goes once the page shows it, or after a minute. We set no advertising or analytics cookies, and our own pages run no advertising or analytics scripts in your browser. The usage records described above go to PostHog from our servers. We do not track you across other websites.
Patchy Cloud does not respond to Do Not Track signals. We do not sell personal information or share it for cross-context behavioral advertising.
Children
Patchy Cloud is a tool for work and is not directed to children. We do not knowingly collect personal information from children.
Changes to this notice
We post each revision here, and it takes effect on the date above. We give further notice, or ask for consent, where the law requires it.
Contact us
Questions about privacy? Email us.