Open Wearables Release 0.5.0: Webhooks, Sleep Fixes, MCP Updates
Key takeaways
- Outgoing webhooks are complete. Register HTTPS endpoints, subscribe to event types, receive normalized payloads in near real-time. Management UI included. Powered by Svix, marked beta.
- Incoming webhooks for Whoop and Suunto. Providers push data to Open Wearables instead of waiting for the next scheduled pull. Requires switching sync mode in admin panel and registering the webhook URL with the provider.
- Sleep calculations significantly more accurate: UTC grouping bugs, awake time inclusion, and duplication during merge all fixed.
- HRV-CV now includes a normalized 0-100 score. MCP Server gets a new get_timeseries tool for granular sample queries.
Open Wearables 0.5.0 is available on GitHub.
Webhooks
Open Wearables has two sync modes. The default is periodic pull: the platform fetches new data from each provider on a configured schedule. 0.5.0 adds the second mode: webhook-based sync.
Outgoing webhooks
Register one or more HTTPS endpoints in the dashboard and subscribe them to event types. When matching data arrives, Open Wearables delivers a POST with the full normalized payload. Delivery runs in the background so it does not block data ingestion. The management UI shows delivery history and lets you retry failed deliveries without touching config files.
Incoming webhooks: Whoop and Suunto
Instead of waiting for the next scheduled pull, Whoop and Suunto can now push data to Open Wearables as soon as it is available. Latency drops from hours to seconds or minutes. Enabling this requires two steps: switch the sync mode in the Open Wearables admin panel, then register the webhook URL in your developer account on the provider side.
Sleep calculation fixes
Three bugs fixed: UTC grouping that incorrectly split sessions across midnight boundaries, awake time incorrectly included in total sleep duration, and record duplication when merging data from multiple sources. All fixes apply to new calculations going forward.
HRV-CV score
HRV-CV now returns a normalized 0-100 score alongside the raw coefficient of variation value. Storage precision improved.
MCP Server: get_timeseries
New tool that returns raw time-series samples rather than aggregated session summaries. Useful for AI features that need to analyze patterns within a session, not just session-level stats.
Want to see it in action? Book a demo
