Stop hand-rolling cooldowns, blocklists, and lease logic. Grab the healthiest resource and report what happened — a verified open-source engine keeps the pool healthy for you.
Powered by a fully open-source, formally-tested engine
// healthiest resource, right nowvar lease = advisor.acquire("checkout-us");useResource(lease.resource()); // report the outcome — that's itadvisor.report(lease, Outcome.success(latency)); // on failure? Outcome.failure(TIMEOUT)// → engine cools, isolates & re-probesJust report what happened. When a resource keeps failing, the engine benches it for a cooldown and takes it out of rotation — then eases it back in over time to check whether it's healthy again. So a bad proxy fixes itself instead of ruining your run.
Reputation is tracked per resource × context — its own cell. If checkout-us goes bad on a proxy, search-eu on that same proxy is untouched — each context keeps its own health and cooldown.
Pool status, reputation curves, and a live event stream — at a glance, with no SQL and no guessing. Metrics also ship to /actuator/prometheus, straight into your own Grafana.
No add-ons, no paywalled basics — everything a serious workload needs is here from day one.
One engine, two ways in — gRPC for the fast path, REST for your tooling.
Subscribe to every hand-out, cooldown, block, and recovery the moment it happens.
Scrape /actuator/prometheus straight into your own Grafana.
Every cooldown, block, and hand-out decision saved to a durable, queryable log.
Health tracked per resource × context — one context's failures never spill into another.
The core is Apache-2.0 on GitHub. Self-host it, or let us run the hosted API.
Works over gRPC or REST — register your resources once, then grab one and report the result per context. Cooldown, isolation, and recovery are never your code to write.
Create an API key in the dashboard and point your client at the gRPC or REST endpoint.
rp_live_…Register your resources once, then ask for the healthiest one per context.
acquire("checkout-us")Tell the pool what happened. Cooldown, isolation, and recovery all happen automatically.
report(lease, success)Issue a key, register a resource, then acquire and report — the quickstart walks the full round trip, in your language.
Key → Register → Acquire → Report, end to end, copy-paste ready.
Start building →ReferenceSix gRPC RPCs and the REST control plane, every field documented.
Browse the API →GuideReputation, cooldown, contexts, leases — how the engine actually thinks.
Learn the model →The engine that makes every one of these decisions is open source. Read the code on GitHub →
No self-serve signup yet — we set up each team by hand so your pool starts healthy. Tell us about your workload and we'll get you a key.
or write to digle117@gmail.com