Service Level Agreement

This SLA applies exclusively to customers with an active paid subscription on the Scale plan. Free and Pro plans are provided on a best-effort basis and are not covered by any contractual availability commitment.

The SLA covers the core event ingestion API at https://api.recalled.dev/v1/events, the query API at /v1/events (read), and the dashboard authentication endpoint. Features marked as beta or preview, the embeddable UI React component, third-party integrations, and the documentation website are not covered.

Recalled commits to a monthly uptime of 99.9% for the covered endpoints on the Scale plan. Monthly uptime is calculated over a full calendar month in the Europe/Paris timezone.

99.9% monthly uptime corresponds to a maximum of approximately 43 minutes and 12 seconds of downtime per month.

Uptime is continuously published on our public status page at https://recalled.openstatus.dev, which is the authoritative source for measured availability.

Uptime is measured by synthetic probes running every 60 seconds against the covered endpoints from at least one location in the European Union, as published at https://recalled.openstatus.dev.

A minute is counted as "unavailable" if at least two consecutive probes fail with an HTTP 5xx response or a timeout exceeding 10 seconds. Intermittent probe failures lasting less than two consecutive minutes are not counted as downtime.

Partial outages affecting less than 50% of requests over a rolling 5-minute window are not counted as downtime.

The data published on the status page is the only valid reference for measuring compliance with this SLA. Customer-side monitoring is not accepted as primary evidence.

The following events are excluded from downtime calculation and do not count against the monthly uptime target:

(a) Scheduled maintenance announced at least 48 hours in advance via email to account owners, capped at 4 hours per calendar month and performed outside business hours (Europe/Paris) when practical.

(b) Force majeure events, including but not limited to natural disasters, war, terrorist actions, labor strikes, regional internet backbone failures, and denial-of-service attacks of a magnitude that cannot be mitigated by industry-standard measures.

(c) Failures caused by third-party infrastructure beyond Recalled's reasonable control, including but not limited to cloud or hosting provider outages, DNS provider failures, certificate authority failures, and upstream network outages.

(d) Failures caused by the customer, including exceeded plan quotas, invalid or revoked API keys, rate limiting triggered by abnormal traffic patterns, incorrect client-side configuration, or use of the Service in violation of the Terms of Service.

(e) Continuous outages lasting less than 10 minutes, which are treated as transient and not counted against the monthly target.

(f) Features and endpoints explicitly marked as beta, experimental, or preview.

The official Recalled SDK for Node.js and browsers (npm package "recalled") includes built-in resilience mechanisms designed to protect your audit trail during transient API unavailability. Events emitted via the SDK while the ingestion API is unreachable are held in an in-memory queue on the client side and automatically retried with exponential backoff for up to 24 hours.

This means that short interruptions of the covered endpoints are transparent to your application: no exception is raised to your code, and events are delivered as soon as the API becomes reachable again. A manual flush() method is also available for short-lived processes that must complete delivery before exiting.

Customers are responsible for integrating the SDK in a way that keeps the client process alive long enough for the retry mechanism to complete. Recalled is not liable for events lost due to premature process termination before the retry window elapses.

In the event of a confirmed breach of the monthly 99.9% availability commitment on the Scale plan, the affected customer is entitled to a service credit applied to the next monthly invoice.

The service credit is calculated as follows: 10% of the monthly fee for the Scale plan for the affected month. The credit is capped at one full month of fees and cannot exceed the amount actually paid for the affected month.

Service credits are the sole and exclusive remedy available for any breach of this SLA. Recalled is not liable for any additional damages, including but not limited to loss of profit, loss of data, business interruption, or reputational damage.

Service credits cannot be exchanged for cash or a refund and do not accumulate across months.

To claim a service credit, the affected customer must send an email to contact@recalled.dev within 30 calendar days following the end of the month in which the breach occurred.

The claim must include: (a) the customer's account email, (b) the covered endpoint(s) affected, (c) the approximate time window of the incident, (d) a brief description of the impact, and (e) any supporting evidence from the customer's own monitoring.

Recalled will review the claim against the data published on the status page. If the breach is confirmed, the service credit is applied to the next monthly invoice within 15 business days of confirmation.

Claims submitted after the 30-day window are automatically denied. Claims submitted without sufficient evidence may be denied or require additional information.

Recalled may update this SLA from time to time. Material changes, including any reduction of the availability commitment or increase of the exclusions, will be notified to active Scale customers by email at least 30 days before they take effect.

Continued use of the Scale plan after the effective date of a revised SLA constitutes acceptance of the updated terms.

For any question about this SLA or to submit a service credit claim, write to contact@recalled.dev. We aim to reply within 2 business days.