Isolated sandbox · your source is never modified

Practice debugging real production failures.

Turn a safe copy of any repository into a focused incident. Investigate the evidence, repair the fault, and get a clear score.

No private repository or API key required.

SEV-2Billing service

14:14 UTC

HTTP 500 errors began shortly after a database migration. Existing customers appear unaffected.

billing-apiDegraded

authenticationHealthy

dashboardHealthy

error  GET /billing → 500   request_id=req_8f2a

From repository to scored rehearsal

The evidence-driven work between “something is broken” and “we know it is safe.”

Identify services, routes, migrations, tests, and reliability risks before selecting an exercise.

Inject a deterministic fault only after the clean baseline passes. Your source stays untouched.

III

Use code, logs, tests, database evidence, and health checks. Then receive a scored report.

Each case has a known root cause, observable evidence, a safe repair path, and deterministic validation.

DB

Unsafe schema transitions, incomplete backfills, and mismatched record creation paths.

CFG

Container host mismatches and environment-specific connection errors.

API

Provider schema drift, partial responses, retry safety, and timeout handling.

SUPPORTED LANGUAGES

TypeScriptJavaScriptPythonJavaCC++C#GoRustRubyPHPKotlinSwiftScala

Temporary copies. Redacted secrets. Allowlisted commands. Expiring workspaces.

Training a team, or hiring? Turn your own incidents into practice. See how →