Game Crash Reporting & Performance Monitoring
Report a crash or a perf sample with one HTTP call. See your errors grouped by fingerprint and your frame-rate / load-time percentiles on a clean dashboard — from any engine, or driven by an AI agent.
Crashes, grouped
POST an error and we fingerprint it — the same crash across thousands of sessions collapses into one group with a live count.
Performance percentiles
POST FPS, load time or any metric and read p50/p95 per build — catch the regression before your players do.
No SDK, any engine
Plain HTTP from Unity, Godot, Unreal, web or a server.
Free to start
Report all the crashes and samples you like. You pay only for active players.
Crash reporting and performance monitoring without an SDK
You do not need a heavy agent to know what is crashing and how your game performs. Backline monitoring is two HTTP endpoints: POST an error (message, type, stack, platform, version) and we group it by fingerprint so a recurring crash is one row with a count, not thousands; POST a perf sample (a metric name and a number, like fps or load_ms) and read p50/p95 percentiles per metric. It works the same from Unity C#, Godot GDScript, a web build, or an AI agent — no engine SDK to import.
Frequently asked questions
How do I add crash reporting to my Unity or Godot game?
Create a free project for an API key, then POST /api/backline/v1/errors with a message, type, stack, platform and version when you catch an exception. GET /errors returns your crashes grouped by fingerprint with counts. Plain HTTP from any engine or an AI agent.
How do I monitor performance / FPS?
POST /api/backline/v1/perf with a metric name and a value (for example {"metric":"fps","value":58} or {"metric":"load_ms","value":1200}). GET /perf/summary returns p50, p95, average and max per metric so you can spot regressions between builds.
Is it free?
Yes. Reporting errors and perf samples is unlimited and free — you pay only for monthly active players (10,000 free, then $0.01 each).
How are crashes grouped?
Each error is fingerprinted from its type, a normalized message (numbers stripped) and the top stack frame, so the same crash across many sessions collapses into one group with a running count and a last-seen time.
Can an AI agent set this up?
Yes — reporting errors and perf samples and reading the groups/percentiles are all documented REST calls an AI coding agent can drive end-to-end.