NetBird
NetBird's second v0.82 candidate keeps the MySQL removal and deprecates Rosenpass.
A side-by-side editorial comparison of Hatchet and Kubeshark — release velocity, themes, recent moves, and the top alternatives to consider.
Hatchet takes durable tasks serverless, with handlers for Cloudflare Workers, Vercel and Node.
Hatchet's new @hatchet-dev/serverless 0.1.0 runs tasks on Cloudflare Workers, Vercel and Node. An in-engine operator delivers them by signed POST or by an invocation websocket that also carries durable sleeps, waits and child runs. Around it, the engine shipped three patches in a day: DAG replays no longer take an advisory lock, worker shutdown now waits for durable tasks, and status queries use partition pruning. The API now declares 429 with Retry-After on every authenticated call.
Kubeshark rebuilt its backend for V2 and is redrawing the free tier around it.
Kubeshark is mid-transition from V1 to V2, announced for January 2026. The v52.11.0 release rewrote the backend entirely — V2 runs on the new implementation while V1 stays on the legacy one — and added cluster-wide L4 PCAP export plus L7-to-L4 mapping that ties API calls back to the TCP and UDP connections carrying them. Since then the releases have been smaller: connectivity map performance, per-flow PCAP files in snapshot archives, and a migration of every service domain from kubehq.com to kubeshark.com.
Hatchet's new @hatchet-dev/serverless 0.1.0 runs tasks on Cloudflare Workers, Vercel and Node. An in-engine operator delivers them by signed POST or by an invocation websocket that also carries durable sleeps, waits and child runs. Around it, the engine shipped three patches in a day: DAG replays no longer take an advisory lock, worker shutdown now waits for durable tasks, and status queries use partition pruning. The API now declares 429 with Retry-After on every authenticated call.
Until now, running Hatchet meant running long-lived workers. The serverless package removes that requirement, so durable execution can run where JavaScript teams already deploy. The engine work points the same way: rate limits declared in the spec, safer worker acks, cheaper status queries. Patch cadence is still very high, often several engine tags a day, and much of it fixes durable-task edge cases.
Expect quick 0.1.x serverless releases as early users hit edge cases on each platform, and possibly matching Python handlers. More engine patches on durable-task shutdown and replay are likely.
Kubeshark is mid-transition from V1 to V2, announced for January 2026. The v52.11.0 release rewrote the backend entirely — V2 runs on the new implementation while V1 stays on the legacy one — and added cluster-wide L4 PCAP export plus L7-to-L4 mapping that ties API calls back to the TCP and UDP connections carrying them. Since then the releases have been smaller: connectivity map performance, per-flow PCAP files in snapshot archives, and a migration of every service domain from kubehq.com to kubeshark.com.
The product is moving down the stack. It started as an API-level traffic viewer and the recent work is all about L4 — connectivity maps, raw PCAP export, mapping L7 calls onto the connections beneath them — which puts it closer to a cluster-wide packet capture tool than a service-mesh observability dashboard. The licensing is moving in the opposite direction of the feature set: an almost-unlimited complimentary licence through the end of 2025 was replaced by one capped at 100 nodes, so the free tier is being defined rather than left open. Two releases also mention agentic infrastructure being staged but not yet usable.
Expect V2 to consolidate onto the new backend and the legacy V1 path to be dropped once the transition completes, since the release notes already describe V1 as being sunset. The staged agentic infrastructure is the most likely candidate for the next substantive feature, given it has been shipped inert twice.
Other Infra & APIs products tracked by Sparkpulse, ranked by recent ship velocity. Each card links to a full editorial trajectory and lets you pivot into a head-to-head comparison with either Hatchet or Kubeshark.
NetBird's second v0.82 candidate keeps the MySQL removal and deprecates Rosenpass.
At Universe, GitHub folds Copilot into Enterprise seats and lets Copilot Fusion pick the model.
incident.io's agent now runs your team's own tools, with a human approval before any change.
Airflow 3.4 reaches its first release candidate, a week after the 10-20 provider wave.
Redocly's reunite.agent gets per-duty cron schedules, rules and source repos.
Jackett ships a daily build to keep hundreds of torrent indexers working as sites move and close.
See all Hatchet alternatives → · See all Kubeshark alternatives →
Latest ship moves from both products, interleaved chronologically. ⚡ = editorial spark.
They serve adjacent needs but don't currently overlap on shipped themes. Hatchet is currently shipping more aggressively (velocity 6.3 vs 0.0), with 0 editorial sparks in the last 30 days against 0. See the at-a-glance table above for a side-by-side breakdown of velocity, recent sparks, and editorial themes.
Sparkpulse doesn't pick a winner — we score release velocity, not feature parity. Hatchet is currently shipping more aggressively (velocity 6.3 vs 0.0), with 0 editorial sparks in the last 30 days against 0. For your specific use case, the alternatives sections above list other Infra & APIs products to evaluate alongside.
Top Hatchet alternatives in Infra & APIs are ranked by recent ship velocity. Browse the "Hatchet alternatives" section above for the current picks, or visit /alternatives/hatchet for the full list with editorial commentary on each.
Top Kubeshark alternatives in Infra & APIs are ranked by recent ship velocity. Browse the "Kubeshark alternatives" section above for the current picks, or visit /alternatives/kubeshark for the full list with editorial commentary on each.