Terms of Service

By creating an account, enrolling an agent with vet-skills join, or using the API, you agree to these terms on behalf of yourself and, if you act for one, your organization. If you do not agree, do not use the service. The public verdict pages and leaderboard may be read without an account and are subject to the sections on acceptable use and disclaimers.

KillerSkills scores agent skills, MCP servers and similar packages by crawling public registries and detonating them in an instrumented sandbox, publishes the resulting verdicts, and lets organizations enforce an install policy on enrolled machines through the vet-skills CLI and hooks. The service is a signal, not a guarantee: a verdict describes behaviour we observed under the conditions on the verdict page, and a skill can change after we score it.

• You must provide a working email address and keep it current. Sign-in is by magic link or an identity provider; there is no password to protect, but anyone with access to your inbox can sign in as you.
• Organization owners and admins are responsible for who they invite, the roles they grant, the policy they publish and the enroll tokens they distribute. Rotate a token you believe has leaked.
• You are responsible for what runs on machines you enrol. The hook and CLI act on your machines with your permissions; removal requests sent from the dashboard delete files on enrolled machines.

• Trial. New organizations get 14 days of the Team plan at no charge. When it ends the dashboard becomes read-only until a plan is purchased; data is retained for 90 days after expiry and may be deleted afterwards.
• Pro. Priced per enrolled person at $12 per seat per month for up to 3 seats, billed monthly in advance in US dollars through Stripe, cancellable at any time with effect from the end of the current month. Pro provides visibility (discovery, blast radius, drift alerts, runtime observation) and does not include enforcement, approvals, remote removal or audit evidence.
• Team. Priced per enrolled person (CI runners excluded) at $24 per seat per month, minimum 10 seats, billed annually in advance in US dollars through Stripe. Seats added mid-term are prorated. Fees are non-refundable except where the law requires otherwise.
• Enterprise. Governed by a signed order form; where it conflicts with these terms, the order form wins.
• Free use. The public verdict pages, the CLI's check command and a daily allowance of on-demand detonations are free for individual use and subject to fair-use limits we may adjust.
• Prices may change with 30 days' notice to organization owners; changes apply at the next renewal.

Do not use the service to attack, probe or overload KillerSkills or anyone else; scrape verdict pages at a rate that affects other users; submit material you do not have the right to submit; circumvent plan limits; or resell verdicts as your own product without a written agreement. Detonation uploads must not contain material that is unlawful to possess. We may suspend accounts that break these rules and will tell you why where we lawfully can.

• You keep ownership of anything you upload or generate: policies, uploaded skills, audit events. You grant us the licence needed to run the service, which includes executing uploaded skills in our sandbox and keeping the artifact to reproduce the verdict.
• Verdicts, scores, findings and the leaderboard are our work product. You may quote them, link to them and embed the badge for your own skills. Bulk redistribution needs permission.
• Publisher claims must be truthful. Claiming a handle you do not control is grounds for removal of the account.
• The vet-skills CLI is open source under its own licence in the repository; these terms cover the hosted service.

A publisher who believes a verdict is wrong can dispute it from the verdict page. We re-detonate, review the findings by hand and either supersede the verdict or explain why it stands, normally within five business days. A dispute does not remove a verdict while it is under review.

How we handle personal data is set out in the privacy policy, which is part of these terms. Our security practices, data residency and how to report a vulnerability are on the security page.

The service is provided as is. We do not warrant that a skill scored as safe is safe, that a blocked skill is malicious, that the sandbox observes every behaviour, or that the service will be uninterrupted. You remain responsible for your own security decisions and for reviewing anything you run.

To the extent the law allows, Cyligence, LLC is not liable for indirect, incidental, special or consequential damages, or for loss of data, profit or business, arising from the service. Our total liability for any claim is limited to the fees you paid us in the twelve months before the claim arose, or one hundred US dollars if you paid nothing. Nothing here limits liability that cannot be limited by law.

You can close an account or organization at any time by emailing [email protected] from the owner's address; paid terms run to the end of the period already billed. We may terminate or suspend for breach of these terms, non-payment, or where continuing would expose us to legal risk, with notice where practical. Sections 6, 9, 10 and 12 survive termination.

These terms are governed by the laws of the State of Delaware, United States, without regard to conflict-of-law rules, and disputes are brought in its state or federal courts. We may update these terms; material changes are emailed to organization owners at least 14 days before they take effect, and continued use after that date is acceptance. Questions: [email protected].