Terms of Service

You must be at least 13 years old to use the Service, or 16 if you live in the European Economic Area or the United Kingdom, and old enough wherever you live to agree to these terms. If you are under the age of majority where you live, a parent or guardian must agree to them on your behalf. You may not use the Service if the law where you live forbids it, or if we have previously closed your account for breaking these terms.

If you accept these terms for a company or other organization, you confirm that you are allowed to bind it, and “you” includes that organization.

Give us a real email address and keep it working: we use it to confirm your account and to send the notices the Service requires. Your username becomes your public handle on the registry. Keep your password and your second sign-in factor to yourself.

Access tokens you create for the command-line tool, the desktop app or the API act as you. You are responsible for what happens under your account and its tokens. Tell us promptly at security@secureagentry.net if you think your account or a token has been compromised. Do not create accounts by automated means, or to get around a limit.

A Business or Enterprise organization is a paid subscription billed per seat through our payment provider, renewing each month until cancelled. Each seat brings a monthly credit allowance into one shared pool that the organization’s members draw on, and an administrator manages the seats. A member belongs to one organization at a time. Joining an organization brings your credit balance into its pool, and its subscription replaces a Pro subscription of your own at once, without a refund for the rest of that cycle.

If a subscription lapses or a payment fails, the organization enters a restricted mode: no credits accrue, and new submissions, hosted runs and added seats are refused, but its people can still sign in, read every QuickAgent, change settings and export the organization’s audit records until it is put right. An organization can be decommissioned by its administrators, or by us where these terms require it. After a notice period this withdraws (yanks) its QuickAgents’ versions, removes its repositories and invitations, and releases its handle and any verified domain.

Where the subscription includes it, an organization’s administrator may take one of the organization’s private QuickAgents off the review pipeline. A version published that way is not scanned: it has no findings, no report and no signature, it is labelled unscanned wherever it is shown, installing it or running it on your own computer asks for consent first, and hosted runs refuse it. The organization takes on the risk of what it publishes that way. This is the one thing a subscription changes about scanning: a public QuickAgent is always scanned, and a QuickAgent cannot be made public while an unscanned version stands.

You keep ownership of what you put into the Service: QuickAgents, repositories, knowledge packs, files you give a run and anything else you upload (“your content”). You confirm that you have the rights to it, and the right to distribute what you publish under the licence you declare.

So that we can run the Service, you grant us a worldwide, non-exclusive, royalty-free licence to host, store, copy, process, scan, analyze, sign, display and distribute your content, to operate the Service and protect it and its users. Scanning includes automated analysis by our own tools and by AI models from third-party providers. For what you publish publicly, the licence also covers making it available to other users to install and run, and keeping every published version unchanged so that installed copies keep verifying. Withdrawing a version (yanking it) stops new installs but does not reach into existing ones. This licence ends when your content is deleted, except for published versions and records that these terms say outlive it.

We do not train AI models on your content, and we do not sell it. AI Review and the Drafter send the content they read to the AI model providers that run them, which process it under their own terms. We look at private content only to operate the Service, to deal with security or abuse, to answer your support request, or where the law requires it.

Every public QuickAgent version is scanned and signed before it is listed. Scanning reduces risk; it does not and cannot prove any QuickAgent safe, and we make no such promise. Scanning, AI Review included, will miss things, and an AI reviewer can misjudge what it reads. A signature shows what was published and by whom; it does not vouch for what the QuickAgent does. Read the security report and the declared capabilities before you install or run anything. Installing and running a QuickAgent is your decision and your responsibility.

Most QuickAgents are written by other users, not by us. Many connect to MCP servers, model providers, websites and other services that we do not run. Those are third parties: their own terms and privacy policies apply to your use of them, and we are not responsible for what they do, what they charge or what they do with data you send them. Any dispute with another user about their content is between you and them.

The command-line tool, the desktop app and the other software we publish come with their own licences, which govern your use of that software.

A QuickAgent you run acts for you, with the access you give it: files, credentials, network and tools. That is true on your own computer and in a hosted run on our infrastructure. You are responsible for what a run you start does, including messages it sends, purchases it makes, data it changes and charges it incurs. Review the capabilities a QuickAgent declares and the consent prompts before you go ahead, and watch what it does.

Unsafe Mode, in the desktop app, lets a local run use the commands a QuickAgent declares even where your computer’s catalogue does not name them. The run’s container, network and consent rules still apply, but it widens what a QuickAgent can do on your computer. Turn it on only for QuickAgents you trust; what happens under it is at your own risk.

You can keep secrets with us, such as API keys for model providers and MCP servers, for use in hosted runs. We keep them encrypted, never show them back to you, and use them only for runs you start. The provider behind a key bills you for its use under its own terms, and you are responsible for those charges. Use only keys you are allowed to use this way, and remove any you no longer want us to hold.

Knowledge packs hold documents you choose to give QuickAgents. Upload only material you have the right to use, and do not upload other people’s personal data unless you are allowed to process it. Packs are sealed (encrypted) at rest. Sealing protects them from casual access; it is not a promise that nothing can go wrong, so keep your own copies.

Parts of the Service use AI models: AI Review, the Drafter that turns a description into a QuickAgent draft, and the models QuickAgents call. AI output can be wrong, incomplete or inappropriate, and it can look confident when it is wrong. Check it before you rely on it, and do not treat it as professional advice. As between you and us, a draft the Drafter makes for you is yours, as far as the law allows; you are responsible for what you do with it, and for publishing it as your own.

Publish, install and run QuickAgents in good faith. The following get content removed and accounts suspended or closed:

• publishing malware, or content designed to exfiltrate data, harvest credentials or take actions a user did not agree to;
• trying to evade or defeat the scanners, AI Review, the gate, quarantine or human review, including by hiding behaviour from them;
• trying to escape a sandbox, reach other people’s runs or data, or use hosted runs for cryptocurrency mining, spam or attacks on others;
• using the Service’s AI features to train or distil other AI models, or reselling access to them;
• impersonating a person or organization, or squatting on names to mislead;
• probing, disrupting or overloading the Service, or getting around rate limits, credit limits or promo-code rules (for example with multiple accounts);
• infringing other people’s rights, or anything unlawful.

Good-faith security research is welcome: report what you find to security@secureagentry.net, avoid harm to other users and their data, and give us a reasonable time to fix it before you disclose it.

Every account has a subscription tier. A tier grants credits each month. Submitting a QuickAgent to the review pipeline, AI Review, hosted runs and sessions, drafts and knowledge builds spend them, as the pricing page describes. Unused credits that a tier or a promo code grants expire when the allowance resets; credits you buy in a pack, and earned credits (see Publisher Earnings below), do not. Credits have no cash value and are never exchanged for money. When an action fails on our side, the credits it held are returned to your balance.

Pro is paid monthly through our payment provider and renews automatically until you cancel, which you can do at any time in your settings. A cancellation takes effect when the current cycle ends; the account then returns to the Standard tier with the Standard allowance and caps. Prices do not include taxes unless they say so. Except where the law requires otherwise, fees already paid are not refunded.

A promo code gives what it says, for the period it states, to an account that is eligible to redeem it. Codes have no cash value, and we may withdraw a code that is being abused. Credits admit a submission; they never change what the scanners look for, or what the gate decides for any public QuickAgent.

We may change allowances, the cost of an action and our prices; for material changes we will post notice on the site before they take effect.

Publisher Earnings lets the publisher of a public QuickAgent, the account that owns its namespace, earn credits when other people run it in Agentry. You join in your settings by accepting this section; an organization joins through one of its administrators. Joining needs a confirmed email address, two-factor authentication and an account in good standing. Earnings start when you join, and nothing from before is paid.

Once you have joined, each hosted run or hosted session of one of your public QuickAgents earns you 25% of the credits its payer is charged for it, a session’s time included, when the payer is on Pro, Promo Pro, Business or Enterprise. Nothing earns from a payer on Standard, from runs on someone’s own computer, from your own runs or your organization’s, from a payer whose use is not metered, from submissions or AI Review, or from Agentry’s built-in agents. The person running a QuickAgent pays the same either way: your share comes out of ours.

Earnings are counted for each half of the month, the 1st to the 15th and the 16th to the end, in UTC, and are paid as credits seven days after that half ends: into your balance, or into an organization’s pool. These earned credits do not expire when the allowance resets, are spent after a tier’s or a promo code’s credits and before credits you bought, move into an organization’s pool with you when you join one, and are lost when the account is closed. Like all credits, earned credits have no cash value and are never exchanged for money.

We may hold your unpaid earnings while we look into how they were earned, and release them afterwards. We may forfeit unpaid earnings that came from abuse of the Service or of these terms, that came from a QuickAgent we quarantine, or that belong to an organization that is decommissioned, and we may reverse a payout of such earnings, which can take your balance below zero. Your settings show each of these with its reason. You can leave Publisher Earnings at any time; what you earned before leaving is still paid.

We may change the share, the schedule or this section, or end Publisher Earnings. For a material change we will post notice on the site before it takes effect, and a change never takes back earnings already paid.

The whole Service is in beta, and some features are marked as previews. They are provided to try out, may change or be withdrawn without notice, and are more likely than the rest to fail or lose data.

If you send us ideas or suggestions, we may use them without owing you anything.

If you believe content on the Service infringes your copyright, write to hello@secureagentry.net with: the work, where the content is on the Service, your contact details, and a statement that you have a good-faith belief the use is not authorized and that your notice is accurate. We may remove or disable the content, tell the person who published it, and close the accounts of people who repeatedly infringe.

We may quarantine, block or remove content, and suspend or close accounts, to protect users and the Service or when these terms are broken, including on reports from others. Where practical we say what happened and why rather than making content vanish silently, and you can write to us about a decision. Trust-and-safety records, such as advisories and abuse reports, outlive the content they concern.

You can delete your account at any time in your settings; the Privacy Policy says what deletion does to published content. When an account is closed, unused credits are lost. The sections that by their nature should survive do survive, including the licence for published versions, the disclaimers, the limits on liability and the indemnity.

The Service is provided as is and as available, without warranties of any kind, express or implied, including warranties of merchantability, fitness for a particular purpose, non-infringement, accuracy and uninterrupted or error-free operation, to the fullest extent the law allows. We do not warrant that any QuickAgent, scan result, AI output or third-party service is safe, accurate or fit for your purpose.

To the fullest extent the law allows, Agentry is not liable for indirect, incidental, special, consequential or punitive damages, or for loss of data, profits, revenue or business, arising from the Service, including from any QuickAgent you install or run, any action an agent takes for you, or any third-party service. Our total liability for all claims about the Service is limited to the greater of the amounts you paid us in the twelve months before the claim and US$100. Some places do not allow these limits; where they do not, they apply only as far as the law permits.

You will defend and indemnify Agentry against claims by others, and the losses and costs that come with them, arising from your content, the runs you start and what they do, or your breach of these terms or of the law.

You may not use the Service in breach of export-control or sanctions laws, or if you are a person those laws prohibit us from dealing with.

These terms are governed by the laws of Canada and of the Canadian province in which Agentry is established, without regard to conflict-of-laws rules. Disputes go to the courts located in Canada, and you and we both accept their jurisdiction. If you are a consumer, nothing here takes away the protection of the mandatory laws of the place where you live, or your right to bring a claim in its courts.

We may change these terms. For a material change we will post notice on the site before it takes effect. The date at the top says which version is current, and your account records the version you agreed to at sign-up. If you keep using the Service after a change takes effect, you accept the new terms; if you do not accept them, stop using the Service and delete your account.

These terms, with the Privacy Policy, are the whole agreement between you and us about the Service. We may send you notices by email to the address on your account or by posting them on the site. If a part of these terms is unenforceable, the rest still applies. Not enforcing a part is not a waiver of it. You may not transfer these terms without our consent; we may transfer them to a successor running the Service. Neither of us is responsible for delays caused by events beyond reasonable control.

Questions about these terms: hello@secureagentry.net. Security reports: security@secureagentry.net.