| Question | Short answer |
|---|---|
| Who is responsible | Merget, Inc. ("Merget", "we"). Contact: support@merget.ai. |
| What this covers | The Merget merge queue: merget.ai, sema.merget.ai, auth.merget.ai, demo.merget.ai, the HTTP API and MCP server, the Claude Code plugin, the CLI and the emails we send. |
| What we collect | Your account and sign-in data, your GitHub connection and the repository data the Service needs, queue and run records, billing records, product analytics and support reports, the website's analytics and session recordings, calls booked through Cal.com, and waitlist entries from earlier versions of the website. Section 2 lists the fields by source. |
| Do we train on your code | Merget may train its own models on repository content, what the Service produces and how it is used, collected from November 17, 2026. It first removes the names, email addresses, logins, account identifiers and credentials it finds, and never takes the stored values of your secrets into training data. Deleting your data takes it out of training data within 30 days; models already trained are kept. Section 3. |
| Who else sees it | The subprocessors in section 4 (Google Cloud, Cloudflare, GitHub, Stripe, PostHog, Cal.com, Slack, Google Workspace), your own GitHub organization, and authorities when the law requires. We do not sell personal information or share it for targeted advertising. |
| How long we keep it | The retention table in section 5. Deleting an organization removes its records from our database at once; repository copies are erased within minutes, and training data within 30 days. |
| Your choices | US state privacy rights (access, correction, deletion, portability), an opt-out of the website's analytics under Privacy choices, Global Privacy Control honored as an opt-out of sale and sharing (which Merget does not do; use Privacy choices to stop website analytics), marketing email opt-out, and a deletion path for analytics data, with or without an account. Section 7. |
1. Who we are and scope
Merget, Inc. operates the Merget merge queue (the "Service"): the Merget GitHub App, the dashboard at sema.merget.ai, sign-in at auth.merget.ai, the HTTP API and MCP server, the Claude Code plugin and the CLI. This Statement also covers the website at merget.ai, the demo at demo.merget.ai, and the emails and text messages we send to run the Service.
The Service is offered only to users located in the United States. This Statement is written for United States law.
Merget decides how and why website, account, billing and support data is used. For the contents of Connected Repositories (code, pull requests, checks, reviews, branch rules) Merget acts on behalf of the Organization that connected them, under the Terms of Service, and uses that content to provide, secure and support the Service; Merget may also use it to train its own models, as section 3 describes under Model training, and decides that use itself. The Organization and its GitHub organization decide who may connect a repository and what Merget may do in it.
2. Information we collect
We collect what the Service needs to work and, on the website, what shows us how visitors find and use it. The sources below name the fields. Where a number is stated, it matches the Service's in-app documentation.
Account and sign-in
When you create an account at auth.merget.ai we keep:
- your email address;
- your password credential, held by Google Identity Platform, never by Merget's own servers;
- your Google or GitHub identity when you sign in with one of them;
- your username, display name and avatar (a gallery avatar, or a photo you upload);
- your second-factor data: an authenticator app secret, a phone number (stored encrypted with a key-management service), a secondary email address, and recovery codes; two-factor sign-in is required for every account;
- sign-in events and the history of changes to your second factor;
- your IP address and device details, used to limit sign-in attempts and to send codes and links;
- sessions, refresh tokens and the record of which browser last confirmed your second factor (asked again every 30 days, per browser);
- the OAuth grants you give to agents: the client, the permissions you ticked, the organization you chose, when you approved it and when it was last used; for an agent key, a hash of the key, its last four characters and its expiry;
- your answers to the optional survey at first sign-in, if you answer it;
- the organizations you belong to and your role in each.
GitHub connection and repositories
When you link GitHub, or an Owner installs the Merget GitHub App, we keep:
- your GitHub user id and login, and the user-to-server token GitHub issues for you, stored encrypted;
- each installation's id, its repository list, your collaborator permissions on them, and organization and team membership where the installation granted it;
- webhook payloads from GitHub (kept 14 days, and the record of each delivery 30 days);
- pull requests, checks, reviews, commit statuses and branch rules of connected repositories;
- the repository contents at the commits Merget analyzes: copies of the repository, the code graphs built from them, and build caches.
The GitHub permissions the App asks for, and why, are listed in the Service's documentation (Install the GitHub App, Permissions). Merget reads a repository only after an Owner connects it and someone turns it on.
Queue and run records
For each connected repository Merget records what it did: findings, merge briefs, prepared fixes, resolutions, conflict excerpts, run reports, timings, token counts and model cost, queue events (a batch forming, a merge landing, a hold), and the log of each tool call an agent makes (the tool's name and its arguments, with a secret's value left out and a submitted patch kept only as a digest; the agent's client; whether it succeeded; and how long it took). These records carry the identifiers of the organization, repository, pull request, run and acting account.
Validation secrets and recipes
Validation secrets and build recipes you store for a repository are kept encrypted ("sealed") and are supplied only to the setup step of Merget's own build. Values are never displayed again after saving. A repository's secrets are deleted the moment the repository is removed from the installation; organization secrets are kept until the organization is deleted. Merget's operational logs record a secret's name, scope, the actor and a keyed fingerprint, never the value. Secret values are never used to train models.
Billing
When an Owner pays for a plan or extra credits we keep: the organization's Stripe customer id, its plan, its spending cap, the credit ledger (each run's charge, the monthly statements and each pull request's receipt), invoices, and the Stripe events Merget stores (stripe_events: Stripe's reports of payments, invoices and subscription changes). The billing email, the billing address and the optional tax id are what an Owner types on Stripe's pages. Payment card data is collected by Stripe only; Merget never sees the card number or keeps any of its details. What Merget sends Stripe about the organization is its name, its identifiers and, through the month, its running count of credits past the bundle, updated every five minutes: never anything about your code or your pull requests.
Analytics and support
The dashboard records how it is used, so the Merget team can see which features people rely on, find errors, answer the problems you report and train Merget's own models (section 3). It records:
- Pages and actions. Each page you open, by its kind, without the names, numbers or search terms in its address. Each action you take, with the number that identifies the repository and with choices and counts, never text from you or your repositories.
- Errors. Errors in the dashboard's own code, with their message and where in the code they happened, and requests to Merget that fail, with their status and error code.
- Session recordings. The pages as the dashboard showed them and what you did on them, so the team can replay a session. Before a recording leaves your browser, the dashboard masks code, diffs, finding text, run reports, graph labels, pull request titles, branch and repository names, people's names and email addresses, and everything you type into a field. What the dashboard draws on a canvas appears blank. Recording starts only after you sign in, and never runs on the pages that complete a sign-in or a GitHub installation.
- Your account and organization. Your account's id, name, username and email address and the Merget products it has, and the identifier and name of the organization you are working in, your role in it and its seat counts. When your account is created, Merget's servers record that it was, and whether you signed in with GitHub.
- Support reports. When you report a problem, the report carries your description, your name and email address, the kind of page you sent it from, and a link to the recording of your session when one is running. With Include page context on, it also carries the page's address without the search terms in it, the organization, the repository's name, the pull request, the run, the time, the dashboard's build and your browser; the dialog shows that context before you send it. Reports and support conversations also reach the Merget team's chat workspace.
The product-analytics and support service receives browser and operating-system details, device type and, where the browser reports it, the device's model, screen and window size, language, time zone, the site that sent you to the dashboard (by its origin only, for example https://github.com) and your browser's IP address, which can indicate an approximate location. It does not receive your sign-in token or cookies from Merget. It keeps its identifiers, and once you sign in your account's and organization's details listed above, in your browser's storage; if that storage is unavailable, they last only while the page is open. The dashboard's analytics start when a page loads, whether or not you are signed in; when you sign in, what they recorded before becomes part of your account's record. The service is PostHog (US Cloud); see section 4.
Website
The website at merget.ai records how visitors find and use it, so the Merget team can see which pages and messages work, which sites and campaigns bring visitors, and where the pages are slow or broken, and so Merget can improve its products and train its own models (section 3). Analytics and session recording start when a page loads. On your first visit a notice says so and names the analytics service; Configure in the notice or Privacy choices in the footer lets you turn either category off for this browser (section 7). The analytics service is PostHog (US Cloud; see section 4), and its requests pass through merget.ai's own server, which removes your cookies from them. For each visit the website records:
- Pages and where you came from. Each page's address and title, the site or link that brought you, the campaign parameters in that link (such as
utm_sourceand advertising click ids), where you first arrived, how long you stayed on each page and how far down it you read. - Device and browser. Browser and operating system with their versions, device type and, where the browser reports it, the device's model, screen and window size, language, time zone and the browser's user agent.
- Approximate location and network. Your country, region, city and postal code, with the city's approximate coordinates, which PostHog looks up from your IP address before it discards the address, and the name and number of the network you connect from (your internet provider, or your company where it runs its own network), which Cloudflare reports to merget.ai's server.
- What you do on the page. Clicks, with the text and link of what you clicked; repeated clicks and clicks that do nothing; text you copy; mouse movement and scrolling, shown as heatmaps; the key sections you reach; clicks on Sign up, Log in, Book a call and the support email address, with where on the page they were; and the steps of a booking (opened, loaded, failed, completed).
- Page speed and errors. How fast each page loads and responds, and errors in the page's own code with where they happened.
- Session recordings. The pages as you saw them and what you did on them, with the page's console messages and the timing of its requests, so the team can replay a visit. Everything you type into a field is hidden before the recording leaves your browser, and the Cal.com booking window is not recorded.
- Identifiers and signals. An analytics ID, anonymous until it is linked to an account or a booking (below), which merget.ai's server keeps in a cookie for a year from your last visit so that your visits can be told apart from other browsers'; whether the browser holds a dashboard session (merget.ai reads only whether the
merget-signed-incookie is present); and whether the browser sends Global Privacy Control.
Linking visits to your account. If usage analytics is on when you follow Sign up or Log in, the website gives the dashboard its analytics ID. Once you sign in, the dashboard links that ID to your account, unless it is already linked to another account. Your earlier visits to the website, with where you first came from, become part of your account's analytics record described under Analytics and support, and so do later visits from that browser, whoever makes them, with their session recordings, for as long as the browser keeps the ID. Signing out of the dashboard does not undo the link. Opting out ends it, and so does deleting merget.ai's cookies and site data; deleting the cookies alone ends it from the next page that reaches merget.ai's server, as the Cookie Policy describes. With usage analytics off, no ID is passed: Sign up and Log in instead tell the dashboard to drop an ID the website handed to that tab earlier. A dashboard tab or booking window that received the ID before you opted out can still link it.
Booking a call. Book a call opens a booking window that Cal.com runs for Merget (section 4). If usage analytics is on when the window opens, the website passes its analytics ID to Cal.com in the window's web address. Cal.com receives the name, email address and answers you give it, and your IP address and browser and device details, and, when you book, it sends the booking to Merget with that ID. Merget adds your name and email address, the booking's id, time and type, and the fact that you booked a call to the analytics record of your visits from that browser, before and after the booking. A booking window first opened with usage analytics off carries no ID, so nothing is linked. One opened before you opted out, even if you close and reopen it on the same page, already carries the ID, and a call booked in it is still linked; reload the page to open one without it.
Earlier versions of the website had a waitlist form, and we keep the entries made in it: what the form asked for, such as your name, email address and organization, and what was recorded with a signup, such as roughly where you were, your time zone, your browser's language, whether you were on a phone, tablet or computer, and the site or link that brought you. Emails you send to support@merget.ai, hello@merget.ai or security@merget.ai are kept as correspondence. The Cookie Policy describes each site's cookies and storage.
Model inputs
In Queue and Autonomous modes Merget may send parts of your code to a model service to resolve a conflict or repair a finding: what the repair agent reads in its checkout of the repository, together with the conflict context. Requests go to Google Vertex AI in Merget's own Google Cloud project; under Google's terms, Google does not use them to train its own models. Advisory mode makes no requests to a model service. Merget's model-service credentials are not available to repository code. From November 17, 2026, Merget may also keep what it sends to the model service and what it returns, for Model training; that, and Merget's use of repository content for training in every mode, is in section 3.
3. How we use information
We use the information above to:
- provide and operate the Service: analyze pull requests, run the merge queue, build and validate changes, show the dashboard, answer the API and MCP server, and send the emails and text messages sign-in needs;
- keep the Service secure and prevent abuse: required two-factor sign-in, attempt limits, token revocation, isolated build environments, and the detection of misuse;
- bill for the Service: meter credits, issue statements and receipts, and run payments through Stripe;
- support you and communicate about the Service: answer reports, send notices about changes to the Service or these documents, and send account notices;
- understand how the dashboard and the website are used, and find their errors and slow pages, through the analytics described in section 2;
- measure Merget's marketing: which sites, links and campaigns bring visitors to the website, and which of those visitors sign up or book a call;
- link a visitor's website visits, before and after they sign in, to their account, and to a call they booked, and prepare for and follow up on booked calls;
- develop, train, evaluate and improve Merget's own machine-learning models, and through them the Service and Merget's other products and services, as described below under Model training;
- meet legal obligations, enforce the Terms of Service and protect Merget, its customers and others.
Merget does not use the personal information it holds for targeted advertising, and does not build profiles of you for that purpose.
Model training
Merget may train its own machine-learning models ("Merget Models", as the Terms of Service define them) and use them in the Service and Merget's other products and services. For this it may use the following, collected on or after November 17, 2026, unless a signed agreement with Merget says otherwise:
- customer content: the code and other contents of connected repositories that Merget reads, in every mode, with their pull requests, checks and reviews;
- what the Service produces and records: findings, merge briefs, conflict resolutions, prepared fixes, run reports, queue and run records, and the log of agent tool calls, with their arguments;
- what Merget sends to its model service and what the model service returns;
- the dashboard's and the website's analytics and session recordings, problem reports and support conversations, described in section 2.
Before any of it enters a training dataset, Merget removes the names, email addresses, GitHub logins, account identifiers and IP addresses it finds in it, and scans it for credentials, such as keys, tokens and passwords committed to code, and removes those it finds. Merget never takes the stored values of validation secrets or GitHub tokens into training data. Merget keeps a separate record of where each item came from, so that it can find and remove it later. Data collected before November 17, 2026 is not used for training.
Merget does not sell or license this data or its training datasets, except as part of a merger, acquisition or sale of assets (section 4), and does not give one customer's content to another. Merget Models are used only in Merget's own products and services, and Merget takes reasonable steps to keep them from reproducing one customer's content in what they produce for another. Merget's model service does not train its own models on this data; when Merget fine-tunes a model, it does so in Merget's own Google Cloud project. Merget has turned off the PostHog setting that lets PostHog train AI on Merget's analytics data; PostHog may still use that data in aggregated or de-identified form to maintain and improve its services (section 4).
When you delete your organization or account, remove a repository from the installation (turning it off is not enough), remove a GitHub account or uninstall the GitHub App, or ask us to delete your data, Merget stops using the data that came from it for training and removes it from its training datasets within 30 days (section 5). Models already trained on it are kept and are not retrained to remove it.
Connecticut law asks us to say whether we collect, use or sell personal data to train large language models. Merget collects and uses personal data to train its models, which can include large language models: the data above can include personal data that remains after the removals described. Merget does not sell personal data for this purpose.
4. How we share information
Subprocessors
Merget uses these providers to run the Service and to develop Merget Models. Each processes data to provide its service to Merget. Some may also use it in limited ways for their own purposes, such as security, fraud prevention, legal compliance and improving their service, as their terms with Merget allow.
| Provider | What it does for Merget | Location |
|---|---|---|
| Google Cloud | Hosting and compute, Cloud SQL (the database), Firestore (account data), Secret Manager and Cloud KMS (sealed secrets and keys), Identity Platform (sign-in, passwords, second factors), Vertex AI (model requests in Queue and Autonomous modes, and any fine-tuning of Merget Models, in a United States region Vertex AI supports for it), and the delivery of sign-in emails and SMS codes | United States (us-east1) |
| Cloudflare | DNS, TLS, delivery and edge protection for merget.ai, sema.merget.ai, auth.merget.ai and demo.merget.ai; the website's server, which sets the website's analytics ID, reports the network a visitor connects from, passes the website's analytics requests to PostHog, and receives Cal.com's notice of each booking (the booker's name, email address and answers), passing the name and email address to PostHog; access control for the demo | United States and global edge |
| GitHub | The repositories, pull requests, checks and reviews the Service works on; sign-in with GitHub | United States |
| Stripe | Payment processing, Checkout, the billing portal, invoices and receipts | United States |
| PostHog (US Cloud) | Product analytics, error tracking and session replay for the dashboard and the website, and the dashboard's support window | United States |
| Cal.com | Scheduling: the booking window on merget.ai and the calls booked in it. Cal.com also receives the browser and device details described under Tracking across other websites | United States |
| Slack | The Merget team's chat workspace, where support reports and conversations also arrive | United States |
| Google Workspace | Email for support@merget.ai, hello@merget.ai and security@merget.ai, and waitlist correspondence | United States |
To be told before Merget adds or replaces a subprocessor, email support@merget.ai with the subject "Subprocessor changes"; Merget tells everyone on that list before the change takes effect.
Your GitHub organization
What Merget posts to GitHub (the Merget check, review comments, prepared fixes, temporary branches and merges) is visible to everyone who can see the repository on GitHub, under GitHub's terms.
Tracking across other websites
Merget runs no advertising, social media or tag-manager script on its sites, and the website's analytics ID is Merget's own: apart from PostHog, it goes only to Merget's dashboard when you sign up or log in, and to Cal.com when you open the booking window, as part of that window's web address, so that a call you book there can be linked to your visits. PostHog processes the website's analytics for Merget, as its service provider, under a data processing agreement that forbids it to sell or share the data or to use it for any purpose other than providing its services to Merget; PostHog may use the data in aggregated or de-identified form to maintain and improve its services.
Two other companies may collect information about your activity over time and across websites while you use Merget's sites. The booking window is Cal.com's page: it loads only when you choose Book a call, and Cal.com runs booking windows like it on many other websites. When you open it, Cal.com receives your IP address, details of your browser and device, and the site you opened it from, and it may set cookies and store data in your browser, as the Cookie Policy describes. Cal.com handles your booking for Merget, as its service provider, under a data processing agreement that is part of its terms with Merget, and may also use booking and device data in limited ways for its own purposes, such as security and preventing abuse. Cal.com says its booking pages use no advertising cookies, that it builds no advertising profiles from booking data, and that it does not sell or share personal information. Merget does not control what Cal.com collects; Cal.com's privacy policy at https://cal.com/privacy describes it. Google runs reCAPTCHA on the sign-in page at auth.merget.ai, to tell people from automated sign-in attempts, under Google's privacy policy.
Other disclosures
We disclose personal information when the law requires it, in response to valid legal process, to protect the rights, property or safety of Merget, its customers or others, and as part of a merger, acquisition or sale of assets, in which case the successor is bound by this Statement for the data it receives. We do not sell personal information, and we do not share it for cross-context behavioral advertising.
5. Retention and deletion
Retention
Deleting an organization removes its settings, repositories, pull requests, runs, findings, queue records and access records from Merget's database at once.
When you delete an organization, or uninstall the Merget GitHub App from an account, the copies of repository code that Merget kept for analysis (repository copies, code graphs and build caches) are erased by a job that runs within minutes; for a deleted organization that includes the accounts it had removed earlier. Merget removes the copies of repositories you remove from the installation, and of accounts you removed without deleting the organization, within 7 days of the removal. Copies of repositories you turn off but keep in the installation stay until you remove the repository, or until Merget's storage-driven cleanup takes them. None of it is reachable through Merget after deletion or removal. A copy Merget took into a training dataset follows the training-data rows below instead.
Everything else Merget removes on a schedule. These bounds apply to the Service's own stores; a copy Merget took into a training dataset follows the training-data rows instead. The table says what Merget removes, from when it counts, and within how long; a daily job carries the removals out, so count on up to one more day:
| What Merget removes | Counted from | Within |
|---|---|---|
| Copies of a repository's code kept for analysis (repository copies, code graphs and build caches) after you remove it from the installation, or remove its account without deleting the organization | the removal | 7 days |
| Run history, findings, queue history and pull request records of an account you removed from the organization or uninstalled the App from | the removal or uninstall | 30 days |
| Run reports, repair details, submitted fixes, resolutions submitted to Merget and conflict excerpts of a closed or merged pull request; its runs and their counts stay | the pull request closing | 90 days |
| Webhook payloads, and the record of each delivery | the delivery | 14 days, and 30 days for the record |
| Records of finished background jobs | the job finishing | 30 days |
| Queue history events | the event | 775 days (a month past the two years the analytics All time range reads) |
| Runs of a closed or merged pull request, with their findings, once the credit meter has rated them | the pull request closing | 775 days |
| The log of agent tool calls | the call | 90 days |
| Cached conflict resolutions nobody reused | the last reuse | 60 days |
| Data in Merget's training datasets that came from an organization or account you deleted, a repository or GitHub account you removed, an installation you uninstalled, or data you asked us to delete (models already trained on it are kept) | the deletion, removal, uninstall or request | 30 days |
| Other data in Merget's training datasets (section 3) | the data's collection | kept while Merget develops its models; no bound today |
| The shared per-file parse cache and definition indexes | the last use by any repository | 30 days |
| Deletion receipts | the deletion completing | 24 hours |
The Stripe events Merget stores (stripe_events: Stripe's reports of payments, invoices and subscription changes) | the event | kept as a financial record; no bound today |
| Waitlist entries from earlier versions of the website (what the form asked for and what was recorded with the signup) | the signup | kept until you ask us to remove them; no bound today |
Turning a repository off keeps its settings, history and copies until you remove it; removing it from the installation starts the 7-day clock. Records of a repository you removed from an account that stays connected are kept, so adding it back restores its settings. An open pull request keeps its reports until it closes.
The credit ledger is kept as a financial record: the credits each run was charged, the monthly statements and a pull request's receipt stay after the run's report is removed, and after the pull request or repository is gone, so a statement always adds up. It records what was charged and why, never the code.
Apart from the removals above, run records are kept until the organization is deleted. A repository's sealed validation secrets are deleted the moment the repository is removed from the installation, or its account is removed, moved to another organization or uninstalled; organization secrets are kept until the organization is deleted.
Account data at auth.merget.ai is kept while the account exists. After an account is deleted, its email address and username are free to use again at once; a record that the account was deleted, holding its id and never its email, is kept for 90 days so that other parts of the Service can finish their cleanup. A deleted organization's address is held for 90 days.
Backups follow their own retention: the database keeps 14 daily backups and 7 days of transaction logs for point-in-time recovery, so a deleted record can remain in a backup for up to 14 days after its deletion.
Product analytics, session recordings and problem reports are kept outside Merget's database, by PostHog, the service described under Analytics and support and under Website in section 2; problem reports and support conversations also reach Slack (section 4); bookings are kept by Cal.com. They are kept as follows; a copy Merget took into a training dataset follows the training-data rows above instead:
| What | Kept |
|---|---|
| Analytics events from the website and the dashboard | 2 years from the event; PostHog then removes them in periodic batches, so some can remain for a short time after |
| Session recordings from the website and the dashboard | 30 days from the recording |
| The analytics record of a person: for a website visitor who follows Sign up, Log in or Book a call, where they first came from (the site, the link and its campaign parameters) and their approximate location; an account's details; and the name and email address from a booking | until you ask us to delete it; no bound today |
| The analytics record of an organization the dashboard was used in: its identifier, its name and its seat counts | until you ask us to delete it; no bound today |
| Problem reports and support conversations, in PostHog and in Slack | until you ask us to delete them; no bound today |
| Bookings made through Cal.com (your name, email address, answers and the call's time) | in Merget's Cal.com account until Merget deletes them; the booking emails Merget receives are kept as correspondence |
Deletion
- Organization. An Owner deletes it in Settings › Danger zone. Deletion affects every member and is permanent. Your GitHub account and GitHub repositories are not deleted.
- Account. Every signed-in member can delete their own account in Settings › Danger zone, after deleting the organizations they own. You are signed out after deletion succeeds.
- One repository or one GitHub account. Turn the repository off, or an Owner removes the account in Settings › GitHub; its repositories are disabled and reset, but the App remains installed on GitHub. Only uninstalling the App on GitHub revokes its access there.
- Analytics and support data. Deleting your account or organization does not remove product analytics, session recordings or problem reports from the analytics service. To have them deleted, write to support@merget.ai.
- Website analytics without an account. Open Privacy choices at the foot of any merget.ai page, copy the analytics ID it shows, and email it to support@merget.ai. We delete the recordings and any analytics record held under that ID, and ask PostHog to delete the events recorded under it; PostHog removes events in weekly batches, so this can take a few weeks. If some events cannot be removed, we tell you, and they go when their retention period above ends. Apart from Merget and the services that process it for Merget (PostHog, and Cal.com if you opened the booking window), only your browser holds the ID, so holding it is how we verify the request. Opting out or clearing your cookies removes the ID from that browser, so copy it first. If your visits are linked to an account, write from the account's address instead.
- Training data. Deleting an organization or account, removing a repository from the installation (turning it off is not enough), removing a GitHub account, uninstalling the App, and each deletion request also remove the data that came from them from Merget's training datasets within 30 days. Models already trained on it are kept (section 3).
- Earlier erasure, backups and other data requests. Write to support@merget.ai.
6. Security
- Two-factor sign-in is required for every account. Agents sign in only through OAuth with your approval and never see your password.
- Every connection uses TLS. Data at rest in Cloud SQL, Firestore and on Merget's disks is encrypted by Google Cloud's default encryption.
- Validation secrets, GitHub tokens and phone numbers are sealed with keys in Cloud KMS.
- Each preparation runs in an isolated environment for that repository. Repository code receives none of Merget's own credentials, and Merget's model-service credentials are not available to repository code.
- Webhooks from GitHub, Stripe and Cal.com are verified by signature before they are acted on.
- Merget's people have the least access needed to run the Service and to train Merget Models, and access is logged.
No system is perfectly secure. If you find a vulnerability, follow the Vulnerability Disclosure Policy at https://merget.ai/legal/vulnerability-disclosure or write to security@merget.ai. If a security incident affects your organization's data, Merget notifies its Owners without undue delay.
7. Your rights and choices
US state privacy rights
Depending on where you live, state privacy law (for example the California Consumer Privacy Act as amended) may give you the right to:
- know what personal information we hold about you and receive a copy of it in a portable form;
- correct inaccurate personal information;
- delete your personal information, subject to the exceptions the law allows (for example financial records we must keep); deletion removes it from Merget's training datasets, but not from models already trained, as section 3 describes;
- opt out of the sale or sharing of personal information and of targeted advertising. Merget does neither, so there is nothing to opt out of; the website's analytics have their own opt-out, described below;
- not be discriminated against for exercising these rights.
You may use an authorized agent to make a request on your behalf; we will ask the agent for proof of authority and may ask you to confirm it.
Global Privacy Control
Merget honors the Global Privacy Control signal. Where a browser sends it, Merget treats it as an opt-out of the sale or sharing of personal information and of targeted advertising. Merget does neither on any of its sites, so the signal has nothing to switch off. The website's analytics and session recordings, which PostHog processes for Merget as its service provider, continue in a browser that sends the signal, and each event records that it did. If your browser sends the signal, Privacy choices confirms that Merget honored it. If Merget ever adds a script or service that sells or shares personal information, it will not run where the signal is present. To stop collection, turn both categories off under Privacy choices.
Do Not Track
Merget does not respond to Do Not Track signals. The website's analytics and session recordings run the same whether or not a browser sends one. To stop them, turn both categories off under Privacy choices; Merget honors Global Privacy Control as described above.
Opting out of the website's analytics
Open Configure in the notice or Privacy choices in the footer. Turn off Usage analytics or Session recording, then select Save choices. Turning both off stops website analytics and session recording in this browser. When both are off, the page deletes the analytics cookie and storage, merget.ai's server deletes the analytics ID and sets no new one while the choice stands, and from then on the website passes no ID to the dashboard or to Cal.com. The choice is kept in a cookie for a year from your last visit and in your browser's storage; while either still holds it, the page records it again in the other, so clearing your cookies alone does not undo an opt-out, except in Safari after 7 days without a visit, when Safari has already erased the storage. Clearing merget.ai's cookies and site data together forgets the choice: the notice appears again and the analytics start as on a first visit, so opt out again afterwards. What was recorded before you opted out is kept on the schedule in section 5 unless you ask us to delete it.
How to exercise your rights
Much of your data is yours to change or delete directly: your profile in Settings › Account, your second factor and sign-in methods on the Merget account page, your agents in Settings › Agents, and your organization or account in Settings › Danger zone.
For anything else, email support@merget.ai from the address on your account, or say which account the request concerns. We verify a request by the account's sign-in or by a reply from the account's address; a request about the website's analytics from someone without an account names the analytics ID shown under Privacy choices, and holding that ID is the verification. We answer within 45 days; where a request is complex we may take 45 more days and will tell you. If we decline a request, you may appeal by replying to our answer, and we will respond to the appeal in writing with the reasons.
Marketing email
Emails about the Service itself (sign-in links, codes, receipts, notices of changes) are part of running the Service and have no unsubscribe. Marketing email, including waitlist updates, carries an unsubscribe link, and we honor a request to stop within 10 business days.
8. Cookies and similar technologies
Merget uses cookies and browser storage only to keep you signed in, to remember your preferences and your privacy choice in this browser, and for the analytics described in section 2, on the website and on the dashboard. There are no advertising cookies. The Cookie Policy at https://merget.ai/legal/cookie-policy lists every cookie and storage key by site, with its purpose and lifetime, and your controls.
9. Children
The Service is for business use and is not directed to anyone under 18. We do not knowingly collect personal information from anyone under 18; if you believe we have, write to support@merget.ai and we will delete it.
10. Changes, prior versions and contact
Changes
We may change this Statement. We post each new version here with its effective date. For a material change we also email account holders at their account address at least 30 days before it takes effect. A change that would use information we already hold in a materially different way applies to that information only with your consent. Non-material changes, such as a clarification or a new hosting or email provider doing the same work, take effect when published; a new analytics or session-recording service is a material change.
Prior versions
This version replaces the version effective November 15, 2026.
Contact
Merget, Inc.
418 Broadway, Unit 7259, Albany, NY 12207
support@merget.ai