VITNATerms of Service
Last updated: October 12, 2026 · COSTRINITY INC. · Regina, Saskatchewan
Under section 13, material changes are posted at least 14 days before they take effect. The sections below show the new text.
Changes to sections 4, 5, 6 and 8 take effect on October 20, 2026. They correct what those sections say the product does: how an account is created, which plans are sold at checkout, what VITNA Desktop and local mode keep, what you can export, and how account deletion works and what it keeps. Section 4 also widens what you are responsible for keeping safe, from your password to every way you sign in: your password, your email account, your GitHub account and your passkeys. Until October 20, 2026, sections 4, 5, 6 and 8 read as follows and remain in force:
4. Your Account (in force until October 20, 2026). Cloud users: you create an account with an email + password. You're responsible for keeping the password safe. If your API key leaks, rotate it immediately from Dashboard → Agent → Rotate Key. We're not liable for events submitted from a compromised key. The rotation flow is the remedy. Desktop users have no account; the API key is generated locally and lives only on your machine.
5. Pricing & Billing (in force until October 20, 2026). Free tier is free, forever. Paid tiers are purchased one month at a time via PayPal (USD, card accepted); INR/Razorpay checkout for India-region accounts is pending merchant onboarding. There is no auto-renewing subscription and no annual billing: each payment grants 30 days of paid access, after which the account returns to the free tier until you purchase again. We don't auto-upgrade you and we don't silently stop accepting your events — if your usage runs well above your tier's guideline volumes we'll contact you. Questions or refund requests: hello@costrinity.xyz.
6. Your Data (in force until October 20, 2026). You own your event data. Cloud storage is the default: your events and decision records are stored on our servers, subject to the retention windows in the Privacy Policy, and we can see what you send us. Local and desktop mode is the opt-in alternative that keeps event payloads on your device. You can export everything as JSON or CSV from the dashboard at any time, and export signed evidence packages via the evidence export endpoint. You can delete your account or wipe local data with a single click. On account deletion we delete your server-side event data within 30 days; append-only audit rows are retained with the owner reference removed. See the Privacy Policy for exactly which mode stores what where.
8. Termination (in force until October 20, 2026). You can delete your account anytime from Dashboard → Settings → Delete Account. We can suspend or terminate your account if you violate the acceptable-use rules in §3, with as much warning as the situation allows (typically 14 days; for severe abuse, immediate). On termination by either party, you keep your local data (it's yours) and we delete the server-side account record within 30 days.
Changes to sections 2 and 3 take effect on October 27, 2026. They describe what VITNA is: a containment layer for AI agents that checks tool calls for the stdio MCP servers you place behind it, holds or blocks risky calls and keeps signed records, and whose compliance checks are decision support, not legal advice. What you may and may not do with VITNA does not change. Until October 27, 2026, sections 2 and 3 read as follows and remain in force:
2. What VITNA Does (in force until October 27, 2026). VITNA is a compliance evidence layer for AI agents. You point your agent's events at our ingest endpoint (the default, stored in our cloud) or at the local SQLite store (opt-in); we render them as a searchable, real-time dashboard with threat detection, severity classification, compliance tagging, and alert routing, and we produce Ed25519-signed evidence packages you can verify offline. The cloud build runs at vitna.costrinity.xyz; the desktop build keeps event payloads on your machine. Detection is heuristic pattern matching, and those limits are documented publicly. Outside guard mode, VITNA evaluates and records; your systems enforce the decision. In guard mode, the npm package sits between an AI agent and the MCP servers you choose to place behind it, and forwards, holds or refuses tool calls to those servers according to your policy. It acts only on the servers you place behind it; VITNA cannot see or act on any other tool, server or route available to your agent, and each session record states which servers were covered. A held call is refused if nobody decides within the window you configure. Choosing what to place behind the guard, setting the policy, and deciding holds in time are your responsibility. Bait, optional. If you turn on honeytools or honeytokens in a guard-mode policy, VITNA adds tools that do nothing, and credentials that open nothing, to what your agent sees from the servers you place behind the guard. A call to a bait tool, or a bait credential reappearing in a later call, is refused and recorded. Bait is off unless you turn it on. You decide where it is appropriate, including anywhere a person might read your agent's tool output, and we do not warrant that bait catches any particular behaviour. See the Privacy Policy for which mode stores what where.
3. Acceptable Use, "Don't use VITNA to attack others" (in force until October 27, 2026). VITNA is the compliance evidence layer for your own agents. Don't pipe events labeled as someone else's agents, or use the platform to facilitate attacks on third parties.
- ›Use VITNA responsibly: no crimes, no abuse, no using us to attack others.
- ›Free tier is genuinely free. Paid plans are bought a month at a time and do not renew automatically.
- ›Your data is yours. Export anytime. Delete anytime.
- ›No measured uptime commitment yet. If we are down, desktop mode keeps working; webhooks are not an independent fallback.
- ›No warranty. Liability capped at what you paid us in the last 12 months. Standard SaaS terms.
1. Who We Are
VITNA is a product of COSTRINITY INC. ("COSTRINITY", "we"). These terms govern your use of vitna.costrinity.xyz, vitna.costrinity.xyz/dashboard, the VITNA Desktop application, and the @costrinity/vitna-compliance-mcp npm package. By using any of these, you agree to these terms.
2. What VITNA Does
VITNA is a containment layer for AI agents. It checks tool calls for the stdio MCP servers you place behind it, holds or blocks risky calls, and keeps signed records of its decisions (signed by VITNA on a claimed account; VITNA Desktop signs them with a key on your machine). Its compliance checks are decision support, not legal advice. You point your agent's events at our ingest endpoint (the default, stored in our cloud) or at the local SQLite store (opt-in); we render them as a searchable, real-time dashboard with threat detection, severity classification, compliance tagging, and alert routing, and we produce Ed25519-signed evidence packages you can verify offline. The cloud build runs at vitna.costrinity.xyz; the desktop build keeps event payloads on your machine. Detection is heuristic pattern matching, and those limits are documented publicly. Outside guard mode, VITNA evaluates and records; your systems enforce the decision. In guard mode, the npm package sits between an AI agent and the stdio MCP servers you choose to place behind it, and forwards, holds or refuses tool calls to those servers according to your policy. It acts only on the servers you place behind it; VITNA cannot see or act on any other tool, server or route available to your agent, and each session record states which servers were covered. A held call is refused if nobody decides within the window you configure. Choosing what to place behind the guard, setting the policy, and deciding holds in time are your responsibility. Bait, optional. If you turn on honeytools or honeytokens in a guard-mode policy, VITNA adds tools that do nothing, and credentials that open nothing, to what your agent sees from the servers you place behind the guard. A call to a bait tool, or a bait credential reappearing in a later call, is refused and recorded. Bait is off unless you turn it on. You decide where it is appropriate, including anywhere a person might read your agent's tool output, and we do not warrant that bait catches any particular behaviour. See the Privacy Policy for which mode stores what where.
3. Acceptable Use
4. Your Account
Cloud users: you create an account by signing up with an email address and a password, or by claiming a trial account created through the setup API, as the MCP package does on first use. We may also create an account for you and hand you a temporary password; it can be used to sign in once, within 48 hours of being issued, and at that first sign-in you must replace it with a password of your own. You're responsible for keeping the ways you sign in safe, such as your password, your email account, your GitHub account and your passkeys. If your API key leaks, rotate it immediately from Dashboard → Agent → Rotate Key. We're not liable for events submitted from a compromised key. The rotation flow is the remedy. Desktop users have no account; the API key is generated locally and lives only on your machine.
5. Pricing & Billing
Free tier is free, forever. Pro and Team are purchased one month at a time via PayPal (USD, card accepted); INR/Razorpay checkout for India-region accounts is pending merchant onboarding. Enterprise is priced by agreement and is not sold at checkout. There is no auto-renewing subscription and no annual billing: each payment grants 30 days of paid access, after which the account returns to the free tier until you purchase again. We don't auto-upgrade you and we don't silently stop accepting your events: if your usage runs well above your tier's guideline volumes we'll contact you. Questions or refund requests: hello@costrinity.xyz.
6. Your Data
You own your event data. Cloud storage is the default: your events and decision records are stored on our servers, subject to the retention windows in the Privacy Policy, and we can see what you send us. VITNA Desktop is the opt-in alternative that keeps events, decisions and holds on your device. Local mode on the ingest API (storage_mode=local) does not keep events off the network: each event is still sent to our API, which checks it and returns it without storing it. Local mode still writes a few records, such as the agent's last-seen time. From the dashboard you can download your agents and their most recent events as a JSON file at any time; the events export endpoint gives one agent's events as JSON or CSV, and the evidence export endpoint gives signed evidence packages. To delete your account, sign in to the dashboard and, from that signed-in browser, send POST /api/owners/delete-account with confirm set to true; or email hello@costrinity.xyz and we will delete it within 30 days. There is no dashboard button for it, and an API key cannot delete an account. A request to the endpoint deletes your server-side event data and your account record when it runs; copies in our hosting and database providers' logs and backups expire on those providers' schedules. Records we keep after deletion include rows in the append-only security and audit log, among them signed decision records and hold excerpts, with the owner id that links them to your account cleared (one kind of row still names your account, as the Privacy Policy explains), and the record of any VITNA Desktop licence key issued to you, with the email address it was issued to. In VITNA Desktop you can clear one agent's events from that agent's page. There is no single button that wipes all local data: what the app keeps stays in its data folder on your device until you delete it. See the Privacy Policy for exactly which mode stores what where.
7. Service Availability
We aim for high availability but do not currently offer a measured uptime commitment or SLA credits: we do not yet run an uptime probe, so there is no measurement we could hold ourselves to. Desktop mode keeps working when the cloud is unreachable, because it reads and writes locally. Alert evaluation currently runs on the same infrastructure as the dashboard, so do not treat webhooks as an independent fallback during a cloud outage. If any of this changes we will say so here.
8. Termination
You can delete your account at any time, as section 6 describes. We can suspend or terminate your account if you violate the acceptable-use rules in §3, with as much warning as the situation allows (typically 14 days; for severe abuse, immediate). On termination by either party, you keep your local data (it's yours) and we delete the server-side account record within 30 days.
9. No Warranty
VITNA is provided "as is." We don't warrant that it will be error-free, that the dashboard will always render every event in a specific order, or that threat detection will catch every malicious input. The threat-detection patterns are best-effort heuristics: they are not a substitute for a security audit, code review, or an actual SOC. Don't use VITNA as your only line of defense against adversarial agents.
10. Limitation of Liability
Our total liability to you for any claim arising from your use of VITNA is capped at the amount you paid us in the 12 months preceding the claim. For free-tier users, that's $0. We're not liable for indirect, incidental, special, consequential, or punitive damages, even if we should have known they were possible. This is the standard SaaS liability cap; if you need higher liability for an enterprise deployment, contact hello@costrinity.xyz to negotiate a Master Services Agreement.
11. Indemnification
You agree to indemnify us against any third-party claim that your use of VITNA caused them harm. For example, if your agent harassed someone via VITNA-mediated event payloads, or you uploaded copyrighted material as a payload. This is the developer's responsibility. VITNA processes the content your agent sends it (event payloads, the text of actions it is asked to check, and in guard mode the parameters of tool calls to the servers you place behind it) in order to evaluate and record them, but it does not review that content for lawfulness, and what your agent sends and does remains yours.
12. Governing Law
COSTRINITY INC. is a Canadian company based in Regina, Saskatchewan. These terms are governed by the laws of Canada and the Province of Saskatchewan, without regard to conflict-of-law principles. Disputes that aren't resolvable by talking to us at hello@costrinity.xyz will be heard in the courts of Saskatchewan, Canada. VITNA is accessible worldwide: if you use it from outside Canada, you agree Canadian law governs our relationship for the purposes of these terms. If you're a consumer in a jurisdiction with mandatory consumer-protection law (EU, UK, US states, Australia, India, and others), nothing here overrides the rights that local law guarantees you.
13. Changes to These Terms
We can update these terms. Material changes (anything that affects your rights, like the liability cap or the data-handling commitments) will be posted here with an updated date at least 14 days before they take effect (we do not currently send change notifications by email). Non-material changes (typo fixes, clarifications) we just edit, and the Last updated date at the top of this page reflects the change.
14. Contact
Questions about these terms, requests for an MSA, security disclosures, or anything else: hello@costrinity.xyz. We aim to respond promptly.