Service Level Agreement

This Service Level Agreement applies to plans that include it — today, the Enterprise plan — and forms part of the Terms and Conditions for those customers. For every other plan it describes what we aim for, not what we owe.

Availability is measured from outside, by the public status page, against the same endpoints anyone can check. We do not grade our own homework.

1. 01Definitions

   • Gateway — The MCP endpoint at spinrun.ai/mcp and its per-team form, through which agents call connected applications. The Gateway is the component this agreement is about.
   • Monthly availability — The percentage of one-minute checks in a calendar month in which the Gateway answered an initialize request successfully, measured by the status page from two locations, one in the European Union and one in the United States. A minute counts as unavailable only when both locations fail.
   • Unavailable — The Gateway returns a server error or no response within 30 seconds to a well-formed, authenticated request. Errors caused by a customer's own credentials, policies, plan limits or a connected application's outage are not unavailability.
   • Scheduled maintenance — Work announced on the status page at least 48 hours in advance, limited to four hours per calendar month, and scheduled outside 08:00–18:00 Central European Time on business days where practicable.
   • S1 incident — The Gateway is unavailable for all customers, or a security incident affecting customer data is confirmed.

2. 02Availability commitment

   For customers on a plan that includes this agreement, we commit to a Monthly availability of the Gateway of at least 99.5%. That allows roughly three hours and forty minutes of unavailability in a thirty-day month.

   Scheduled maintenance, third-party application outages, force majeure, and unavailability caused by the customer's own configuration are excluded from the calculation.

   We intend to offer a 99.9% tier once three consecutive months of published measurement support it, and not before.

   Plan

   Availability target

   Enterprise

   99.5% monthly, with service credits below

   Business

   99.5% monthly, best effort, no credits

   Pro and Free

   Best effort

3. 03Support response

   Support requests go to sales@spinrun.ai, or to the dedicated channel named in an Enterprise agreement. First-response targets are measured during business hours, 09:00–18:00 Central European Time on Romanian business days, except for S1 incidents, which are measured around the clock.

   Severity

   First response

   S1 — Gateway down for everyone, or a confirmed security incident

   4 hours, at any hour (Enterprise); next business day otherwise

   S2 — A feature unusable for one customer with no workaround

   1 business day

   S3 — Degraded, workaround exists

   2 business days

   S4 — Question or request

   3 business days

4. 04Service credits

   If Monthly availability falls below the target for a customer on a plan that includes this agreement, that customer may claim a credit against the next invoice. Credits are the sole remedy for missed availability and are capped at 50% of the monthly fee for the affected month.

   To claim, write to legal@spinrun.ai within 30 days of the end of the affected month, naming the month. We reconcile against the status page's record and apply the credit within one billing cycle.

   Monthly availability

   Credit

   Below 99.5% but at least 99.0%

   10% of the monthly fee

   Below 99.0% but at least 98.0%

   25% of the monthly fee

   Below 98.0%

   50% of the monthly fee

5. 05How it is measured and reported

   The public status page, at status.spinrun.ai, runs synthetic checks against the Gateway and against the health endpoints of each host every minute from two locations, one in the European Union and one in the United States, and records every incident with its start, end and cause. The monthly availability figure on that page is the one this agreement uses.

   Customers on an Enterprise plan receive a monthly availability report by email; any customer may read the same numbers on the status page at any time.

6. 06Incident communication

   S1 incidents are posted on the status page within 30 minutes of confirmation and updated at least hourly until resolved. Where an incident is also a security incident affecting customer data, the notice obligations in the Data Processing Agreement apply in addition, with their own clocks.

   A post-incident summary is published on the status page within five business days of resolution for any incident that counted against availability.

   Data Processing AgreementSecurity practices

7. 07Changes

   We may revise this agreement with 30 days' notice to affected customers. A revision never reduces a commitment for a customer's current paid term.

   Terms and Conditions