Self-hosted error tracking for Rails. Local variables, breadcrumbs, system health snapshots, N+1 detection, swallowed exception tracking — running in your own process. Zero recurring cost.
All three [self-hosted alternatives] had an issue with error backtrace when using Turbo — RED did fix it… solid_errors and Faultline are not very active projects, RED is very active and @AnjanJ is very responsive in fixing issues. So, RED was my final choice.
Hosted error tracking is excellent — and excellent at extracting recurring revenue. RED gives you the same instrumentation, running in your own process.
Every feature opt-in, production-safe, and zero overhead when off.
Most error trackers give you a stack trace. RED gives you the entire local scope at the moment of `raise` — captured via `TracePoint(:raise)` without storing a single Binding object.
filter_parametersEvery codebase has them: exceptions raised, silently rescued, and forgotten. RED uses dual `TracePoint(:raise)` + `TracePoint(:rescue)` to surface them with hourly bucketing and raise/rescue ratios.
Copy the entire error context — backtrace, cause chain, local variables, breadcrumbs, environment, related errors — formatted for Claude, ChatGPT, Cursor, or whatever assistant you use. App frames only. Sensitive data still filtered.
Every error captures a sub-millisecond runtime snapshot. GC stats, memory pressure, file descriptors, TCP states, system load, Puma worker/thread saturation, DB pool stress. No subprocess calls. All under 1ms budget.
Every error records whether it came from production, staging, uat — any name your deploys use. Filter the index by it, see it as a badge and in every notification, and keep staging out of your pager with notification_environments. The same error in two environments is two rows with their own status.
Rack::Attack track rules now name the client — GPTBot, ChatGPT-User, ClaudeBot, Claude Code, PerplexityBot, Bytespider, CCBot and more, told apart from ordinary crawlers like Googlebot. Per-rule AI agent totals and a top-agent column, without blocking or throttling anything.
Dashboard, emails and notification payloads translated — German, Spanish, French, Portuguese, Italian, Polish, Russian, Ukrainian, Japanese, Simplified Chinese. Switch from the navbar. Your app's I18n is never touched.
Circuit breaker + adaptive sampling. When a bad deploy throws 50k errors/min, the gem degrades itself first — exact counts, zero DB amplification.
v0.8.2Auto-groups by stack fingerprint. Counts occurrences instead of spamming you with duplicates.
Cross-error SQL fingerprinting. Auto-extracted table name with eager-loading suggestion.
Resolved errors that recur reopen automatically with a "Reopened" badge. Never miss a regression.
Slack, Email, Discord, PagerDuty, custom webhooks. Per-error cooldown. Severity filters.
v0.5Auto-create on first occurrence. Two-way webhooks with HMAC. Lifecycle sync.
v0.5.8Full root-cause traversal. SocketError caused RuntimeError caused ServiceError. Never guess again.
"New in this release" detection. Stability indicators. Release-over-release deltas. Zero config.
v0.5.10Rank by unique users affected, not occurrence count. 1000 users x 1 hit beats 1 user x 1000.
See which lines were executed in production. Ruby Coverage API, oneshot mode. Zero overhead when off.
v0.5.11Track multiple Rails apps in one dashboard. Auto-registration, data isolation, easy switching.
Passwords, tokens, secrets, API keys auto-scrubbed before storage. Lambda-based custom rules.
72KB vs 676MB B-tree on occurred_at. 70x faster time-range queries on large datasets.
Try the live demo. See all 30+ features in action. No signup. Walk away in 60 seconds — or stay forever.