Privacy Policy
How Ordo Kit handles information about you: what we collect, what we do with it, who else receives it, and how long we keep it.
Last updated:
Who we are #
Ordo Kit is operated by IE Meshcheriakov Artem Andreevich, an
individual entrepreneur, of Moldo Niyaza 2, Osh, 723510, Kyrgyzstan,
who is the controller of the personal information described here.
Ordo Kit runs the website
at https://ordokit.dev, the dashboard, the MCP server at
https://ordokit.dev/mcp, and the hosting for the sites deployed
through them. In this policy, “we” and “us” mean Ordo Kit, and “you”
means anyone who visits the website or uses the service.
Questions about this policy and requests about your information go to support@ordokit.dev.
This policy covers what Ordo Kit itself collects. A site you deploy is yours: what it collects from its own visitors is up to you and your own privacy policy.
In short #
- You sign in with GitHub. We never see your GitHub password, and we do not receive the email address on your GitHub account.
- We read your repository only to check, screen and build it.
- Card details go to our payment provider, Paddle, never to us.
- We do not receive your conversations with an AI agent: a tool call brings us only the tool's name and its arguments.
- Tokens are stored only as hashes.
- No analytics, no advertising, and no third-party trackers.
- We do not sell personal information.
- You can delete your account yourself, under Billing → Danger zone.
What we collect #
Your GitHub account
You sign in with GitHub, and we ask GitHub only for the
read:user permission. GitHub sends back a token, which we
use once to read your profile and then discard; it is not stored. From
the profile we take:
- your numeric GitHub account id, which is what your account is keyed on;
- your login (username), refreshed each time you sign in;
- your display name and the address of your avatar image.
Your sign-in cookie carries these, together with a GitHub no-reply
address built from your id and login
(id+login@users.noreply.github.com),
which pre-fills the Support form. Our records are keyed on your id and labelled with your
login. Your name and avatar address are not stored in our database.
Repositories
When you choose a repository, we read it to:
- check that you can access it, and list its branches;
- find its Dockerfile and the port it listens on, from the Dockerfile and the file listing;
- screen it for abuse (such as crypto miners, reverse shells and attempts to reach the host) and for leaked secrets. We download the repository's archive into a temporary directory on our server, scan it there, and delete it when the scan ends. The scanners run on our server: your code is not sent anywhere else to be screened;
- build it. Our hosting layer clones the repository on the server that will run it and builds a container image there. The image holds whatever your Dockerfile copies into it, and stays on that server while the project exists.
We use repository contents for nothing else.
Public repositories are read without credentials. Private repositories are read through our GitHub App, which you install on GitHub and choose repositories for. Each time we read one, we ask GitHub for a short-lived token limited to that single repository, with read-only access to its contents and metadata. These tokens are kept in memory only. When you install the app, GitHub also gives us a token for your GitHub account. We use it once, to confirm the installation is yours, and then revoke it at GitHub. We record which installations belong to your account (the installation id, and the name and type of the GitHub account it is on).
To build a private repository we generate a key pair for the project. You add its public half to the repository on GitHub as a read-only deploy key. The private half goes to our hosting layer, which uses it to clone the repository on each rebuild. We keep our own copy only until the project is created (or, if it never is, until you delete your account). The hosting layer's copy is deleted when you terminate the project.
With the GitHub App installed, GitHub notifies us when you push, and we rebuild the projects built from that repository and branch. For each notification we record its id, its type, the repository, and the GitHub login that sent it.
About the repository itself we record its address, the branch, the Dockerfile path, the build context and the port. We also record the screening result: the rule, severity and file path of each finding, never the code or the text that matched.
Projects and hosting
For each project we record its address
(<alias>.ordokit.dev), its owner, the repository
details above, your own domain if you add one, an optional contact
email address, the server it runs on, and its state. The contact
address is where we send the “your site is live” message and
bandwidth notices.
- On our managed server, your container runs on a
server we operate. Traffic to ordokit.dev and to
<alias>.ordokit.devpasses through Cloudflare's proxy. Traffic to your own domain comes straight to our server. - On a server you bring, you give us its address, SSH user and port, and a private SSH key with root access. The key goes to our hosting layer, which uses it to manage the server. It is never written to our own disk, and the hosting layer deletes it when the last project on that server is terminated. A monitoring agent installed on the server reports its running containers to our hosting layer every 60 seconds while a project runs there. We record which version of the access terms you agreed to, and when.
Your site's visitors. We add no analytics or tracking to your site. To meter bandwidth and notice traffic floods we count the bytes and requests of each site, as totals, with no visitor addresses and no pages. Our proxy also limits how fast any one address can send requests; it keeps those counts in memory and does not store them.
Plans and payments
Payments are handled by Paddle (Paddle.com Market Ltd), our reseller and merchant of record, in its own secure checkout, which opens over a page of ours. Your card details and billing details go to Paddle, never to us. When you start a checkout, or a card check for the free trial, we send Paddle your GitHub id and login, so the payment can be matched to your account; you enter your email address on their form.
From Paddle we keep your plan, its status and billing period, and their identifiers for your subscription and for you as a customer. For a free-trial card check we keep the result and the same two identifiers; never the card.
Agents and the MCP server
When you connect an AI agent, such as Claude, to the MCP server, you approve it on our consent page. We record:
- the app's registration: the name it gives itself and the addresses it may send you back to;
- your approval: which app, your login, when you approved it and when it was last used;
- its tokens: an access token that lasts 1 hour and a refresh token that lasts 60 days, stored only as SHA-256 hashes.
API tokens you create for scripts are stored the same way, as hashes, with the name you gave them and when they were created and last used.
We do not receive your conversation with the agent. A tool call brings us only the tool's name and its arguments (such as a repository or an address), which we handle as if you had entered them in the dashboard. For each call we log the tool's name, the names of its arguments (not their values) and your login. When an agent connects, we also log the name it gives for itself. What a tool returns goes to your agent, and from there to the agent's provider, under that provider's own privacy policy.
Support and email
The Support form sends your message by email to our support mailbox. The email contains the name and address you type, your message, and your GitHub login and id. The message is not stored in our database: we record the time it was sent, to limit how many can be sent, and an audit entry with the reply address and the message's length. The email itself stays in our support mailbox; you can ask us to delete it.
We send email through Resend: the “your site is live” message and bandwidth notices to your project's contact address, Support messages to us, and operator alerts to us.
Logs and records
Running the service leaves these records:
- Deployment events: each step of a deployment or teardown, what it was asked to build, the screening findings (rule, severity, path), and failure reasons, which can quote the last lines of build output.
- Activity trail: things you did that leave no other record, such as port detections, screening results, access checks, deployments refused before they started, and requests refused by a rate limit.
- Audit trail: who did what, recorded before it is attempted. Plan changes, tokens created and revoked, Support messages, account deletion, and every action an operator takes on someone else's project.
- Application logs: our service's own log, which names logins, project addresses and repositories as it works. Tokens, keys and passwords are redacted from every line before it is written. Separately, the web server prints one line per request (time, path, status, referring page and browser) to the service's console output. That output is not copied into the log files, and it is discarded when the service is replaced by a later release. The address on those lines is our own proxy's, not yours.
- Bandwidth: bytes sent and received per project and per account in each billing cycle, built from samples taken every few minutes.
- Traffic: the latest request count of each running site and whether it is above the alert threshold. No history is kept.
- Rate limits: counts of recent requests, keyed by your account id or by a hash of your IP address, never the address itself.
- Operator alerts: a record of each alert sent to the people who run Ordo Kit (see Who else receives it).
The people who run Ordo Kit (operators) can see every account's projects, deployment events, activity, audit trail and logs, so that they can support customers and keep the service safe. Operators are not shown your push-to-deploy webhook secret.
How we use it #
- To sign you in and keep your projects tied to your account.
- To check, screen, build, run and remove your projects.
- To apply your plan's limits, and to bill you through Paddle.
- To send the emails described above.
- To answer your support requests and investigate problems.
- To protect the service and its users: screening repositories, rate limits, preventing repeat free trials, and alerting operators to failures, payments and unusual traffic.
We do not sell personal information, and we do not use it for advertising.
Where it is stored #
Our records are kept in a database on our own server, with our log files beside it. Your projects run on our managed server or on the server you bring. What reaches other companies is listed in the next section, and what stays in your browser in Cookies and browser storage.
How long we keep it #
| Record | Kept for |
|---|---|
| Your plan, card check, GitHub installations, API tokens (revoked ones too) and connected apps | While your account exists. A connected app you disconnect is deleted 30 days later. |
| A project's record | Until you terminate the project. A project that failed to deploy keeps its record, marked failed, until you delete your account. |
| Your project's container, data volumes and configuration on its server | Until you terminate the project. |
| Completion details (address, certificate status, push-to-deploy secret) | 24 hours |
| Deployment events | 30 days |
| Activity trail | 30 days |
| Application log files | 30 days after a file was last written to |
| Web server request lines | Until the service is replaced by a later release |
| Operator alert records | 30 days. Copies already delivered stay in the operators' Telegram chat and mailbox. |
| OAuth sign-in codes | Usable for 60 seconds, deleted a day after they expire |
| OAuth access tokens | 1 hour, then deleted |
| OAuth refresh tokens | 60 days, then deleted. Each use replaces the token with a new one. |
| App registrations nobody approves | 24 hours |
| Bandwidth samples | 48 hours |
| Bandwidth totals per billing cycle | 400 days |
| Traffic counts | Until the site stops running |
| Rate-limit counters | Deleted after they expire. No limit looks back more than one day. |
| Times of Support messages | Until they no longer count toward the limit and you send another message, or until you delete your account |
| Progress of an account deletion | 24 hours after it finishes |
| Audit trail | Kept. It is not deleted with your account. |
| GitHub push and installation notifications | 30 days, whether or not you delete your account in the meantime |
| Record of a deleted account (GitHub id, login and date) | Kept, so that a deleted account does not get a second free trial |
Some of these are removed by a clean-up that runs when the service starts, which happens with every release. A record can therefore outlast its period by the time until the next release.
Cookies and browser storage #
We set one cookie, __Host-session, when you sign in. It
holds your sign-in (the profile details listed above and when you
signed in) and short-lived state for the sign-in itself. It is signed
so it cannot be altered, it cannot be read by scripts on the page
(HttpOnly), it is sent only over HTTPS and only to this
site, and it lasts until you sign out or close your browser.
The dashboard also keeps a few things in your browser's storage:
theme(local storage): light, dark or automatic.ordo.projects.<login>(local storage): a copy of your project list, so the page opens quickly. Removed when you sign out.ordo.ghConnectLater(local storage): that you chose “Not now” when offered the GitHub App.ordo.agentTab(local storage): which agent's connection instructions you last viewed.ordo.buy(session storage): the plan you chose before signing in, so the checkout can open afterwards.ordo.pendingRepository(session storage): the repository you were entering while you installed the GitHub App.
There are no analytics, advertising or third-party trackers, and our pages load no scripts, fonts or images from other sites.
Deleting your account #
Go to Billing → Danger zone → Delete account. Deleting your account:
- terminates every project, which removes its container, data and configuration from its server;
- cancels your plan and any free-trial card check with Paddle;
- deletes your plan and card-check records, API tokens, connected apps, GitHub installation records, deploy keys, activity trail, bandwidth records, completion details, and your projects' records and deployment events;
- signs you out everywhere.
What is kept:
- the audit trail, including a record of the deletion that gives only counts of what was removed;
- a record of your GitHub id and login and when the account was deleted, so that signing in again does not start a second free trial;
- log files, operator alert records and GitHub notification records, until their periods above end;
- Paddle's own records of your payments, which they keep under their own policy.
Our GitHub App stays installed until you remove it on GitHub (Settings → Applications). If you added a deploy key to a repository, remove it in that repository's settings as well.
Your choices and requests #
The dashboard shows your projects, plan, API tokens and connected apps. You can revoke a token or disconnect an app there at any time, and delete your account as described above. Your login and name come from GitHub: change them on GitHub, and your login is updated here the next time you sign in.
To ask for a copy of the information we hold about you, to have something corrected or deleted, or to object to how we use it, email support@ordokit.dev. There is no self-service export: we handle these requests by email. We may ask you to confirm that you control the GitHub account concerned before we act.
Depending on where you live, you may have further rights under data protection law, including the right to complain to your data protection authority.
Security #
- The website, dashboard and API are served only over HTTPS.
- API tokens and OAuth codes and tokens are stored only as SHA-256 hashes.
- Sign-in asks GitHub only for
read:user, and that token is not kept. Repository tokens are limited to one repository and to read-only access, and are held in memory only. The token GitHub gives us when you install our app is revoked as soon as it has been used. - Deploy keys are per project and read-only. SSH keys are handed to the hosting layer without being written to our disk.
- Tokens, keys and passwords are redacted from our logs, and the push-to-deploy secret is shown only to the project's owner.
- Repositories are screened before anything is built. Each project runs in its own container with memory and CPU limits.
No system is perfectly secure. If you find a vulnerability, please tell us at support@ordokit.dev.
Children #
Ordo Kit is a service for developers and is not directed at children. We do not knowingly collect information from children. If you believe a child has given us information, contact us and we will delete it.
Changes to this policy #
When we change this policy, we update this page and the “Last updated” date at the top.
Contact #
Ordo Kit (operated by IE Meshcheriakov Artem Andreevich, an individual entrepreneur)
Moldo Niyaza 2, Osh, 723510, Kyrgyzstan
Telephone: +996 755150546
Email: support@ordokit.dev