Prove your RPC endpoints work. Before every upgrade.
Automated regression testing for blockchain infrastructure. Catch latency regressions, correctness issues, and client divergences before they reach production.
Your monitoring tells you something broke.
OnchainProbe proves things work.
Client upgrades silently break RPC behavior
A geth patch changes gas estimation. Your CI is green. Your users find out first.
Multi-provider setups diverge without warning
Different clients return different results for the same call. Nobody notices until production.
Ad-hoc scripts rot and miss edge cases
That curl script from 2023 tests 3 methods. Your endpoint supports 50.
Everything you need to test RPC endpoints
From cross-client comparison to regression detection, in your CI or on a schedule.
Regression Testing
Run endpoints side by side. See exactly what changed in every RPC method.
Cross-Client Comparison
Compare geth, besu, erigon, nethermind. Field by field, across every endpoint.
Test Assertions & Verdicts
Clear verdicts with detailed assertion results per endpoint.
Latency & Performance
P50/P95/P99 latency, per-method bar charts, comparison verdicts.
How it works
Connect your RPC endpoints
Add any EVM-compatible endpoint. Supports auth headers, bearer tokens, and query params.
Pick the test suites you need
Choose from latency, correctness, comparison, and contract deployment suites.
Run on schedule, on blocks, or from CI
Interval scheduling, block-triggered execution, or API-driven from your CI pipeline.
Get verdicts, comparisons, and alerts
Clear verdicts with detailed field-by-field diffs and configurable alerts.
Why not just monitoring?
Monitoring watches for failures in production. It tells you something is wrong after your users are already affected. That is table stakes, not a testing strategy.
Proactive testing is different. It proves your endpoints return the right data, at the right speed, before you deploy. It catches the geth patch that changes gas estimation, the besu update that drops a trace field, the provider migration that adds 200ms of latency.
What teams try instead
✗Curl scripts
Quick to write, impossible to maintain. They test one method at a time, skip assertion logic, and never cover cross-client divergence.
✗Generic API testing
Built for REST endpoints, not JSON-RPC. No concept of execution clients, volatile methods, block-level state, or contract deployment workflows.
✗Uptime monitoring
Confirms "the endpoint responds." Says nothing about whether the response is correct, consistent across clients, or within your latency budget.
OnchainProbe fills the gap between "endpoint is up" and "endpoint is correct."
Proof, not just pings.
Start testing your endpoints in 10 minutes
Pro features free during early access. No credit card required.