Eve library

phase-149.md

Tracker

Revision 1

On this page

Historical task source. The Eve task board holds current task state.

Phase 149: Sovereign Crash Analytics & Error Tracking Platform — Sentry / BugSnag / Rollbar / Crashlytics / Raygun / Honeybadger / Datadog Error Tracking Replacement#

Closes the error-tracking sovereignty gap. Existing phases reference Sentry as a dependency across services (Phase 3, Phase 22, Phase 30–33, others). Phase 98 covers AI observability. Phase 145 CI/CD collects build telemetry. None of these deliver a dedicated multi-language, multi-platform crash ingestion, symbolication, grouping, release-health, and distributed-tracing platform capable of replacing Sentry across Oshun's 45+ domain apps.

This phase delivers a complete sovereign replacement for Sentry, BugSnag, Rollbar, Firebase Crashlytics, Raygun, Honeybadger, Airbrake, and the error-tracking portions of Datadog / New Relic / Instabug.


149.1 @neith/crash-ingestion — Event Ingestion & SDK#

  • 149.1.1 JavaScript / TypeScript browser SDK (source-map aware)
  • 149.1.2 Node.js server SDK with Express / Fastify / Nest middleware
  • 149.1.3 iOS / macOS Swift SDK with crash-reporter bridge
  • 149.1.4 Android Kotlin / Java SDK with NDK native bridge
  • 149.1.5 Rust SDK with backtrace + panic hook
  • 149.1.6 Go SDK with panic / error wrapping
  • 149.1.7 Python SDK with Sentry-API-compat surface
  • 149.1.8 C / C++ SDK with breakpad / crashpad integration
  • 149.1.9 Unity / Neith Engine plugin
  • 149.1.10 Unreal / Neith Engine plugin
  • 149.1.11 Flutter / Dart SDK
  • 149.1.12 React Native / Expo SDK
  • 149.1.13 Electron main + renderer SDK
  • 149.1.14 Kubernetes sidecar ingest agent
  • 149.1.15 OTLP bridge (accept Sentry-envelope + OTLP traces)
  • 149.1.16 Sentry DSN-compat endpoint for drop-in replacement
  • 149.1.17 Rate-limited client buffering with exponential backoff
  • 149.1.18 Offline queue with journalled replay
  • 149.1.19 PII scrubbing at SDK layer (client-side strip)
  • 149.1.20 SDK auto-update channel

149.2 @neith/crash-symbolicator — Symbolication & Deminification#

  • 149.2.1 iOS dSYM symbolication
  • 149.2.2 Android ProGuard / R8 mapping deobfuscation
  • 149.2.3 Android NDK breakpad symbol upload
  • 149.2.4 JavaScript source-map resolver with chain walking
  • 149.2.5 Windows PDB symbolication
  • 149.2.6 Linux DWARF + debug-link symbol lookup
  • 149.2.7 Rust / Go / Swift native stack trace demangling
  • 149.2.8 WebAssembly name-section + DWARF
  • 149.2.9 Unity IL2CPP line.cpp translation
  • 149.2.10 Unreal VS / AS profile symbolication
  • 149.2.11 Symbol server (S3-compat) with CI auto-upload
  • 149.2.12 Symbol server retention by release cohort
  • 149.2.13 Reprocess events when late-uploaded symbols arrive
  • 149.2.14 Source-context lookup (show ±5 lines of code in trace)
  • 149.2.15 Git blame integration per frame (culprit commit)

149.3 @neith/crash-grouping — Fingerprinting, Grouping & Dedup#

  • 149.3.1 Stack-trace fingerprint (normalized frames)
  • 149.3.2 Exception-type + message fuzzy grouping
  • 149.3.3 Hierarchical grouping (loose / strict / custom)
  • 149.3.4 User-defined fingerprint rules (regex / wildcard)
  • 149.3.5 ML-based grouping (embedding clustering)
  • 149.3.6 Split group (over-grouping correction)
  • 149.3.7 Merge groups (under-grouping correction)
  • 149.3.8 Event-to-issue assignment with atomic dedupe
  • 149.3.9 Issue state machine (unresolved / resolved / ignored / reopened)
  • 149.3.10 Regression detection (resolved → reopened)
  • 149.3.11 Auto-resolve when next release ships
  • 149.3.12 Snooze until next occurrence / count / time
  • 149.3.13 Issue merge / unmerge from UI
  • 149.3.14 Release-specific grouping scope
  • 149.3.15 Cross-project issue link (same bug across services)

149.4 @neith/crash-storage — Event Storage & Query#

  • 149.4.1 ClickHouse-class column store for events (≥ 10M / day)
  • 149.4.2 S3-compat blob store for full event payload
  • 149.4.3 Retention by plan / project (7 / 30 / 90 / 365 d)
  • 149.4.4 Sampling at ingest with deterministic hash
  • 149.4.5 Search by user / release / tag / env / URL / custom
  • 149.4.6 Full-text search on message + trace
  • 149.4.7 Saved search with notification rule
  • 149.4.8 Query builder UI (drag-drop filters)
  • 149.4.9 SQL-like query API for advanced users
  • 149.4.10 Aggregate views (count per group per hour)

149.5 @neith/crash-releases — Release Health & Tracking#

  • 149.5.1 Release entity with commit-range + version
  • 149.5.2 Deploy event with environment + timestamp
  • 149.5.3 Crash-free-session rate per release
  • 149.5.4 Crash-free-user rate per release
  • 149.5.5 Adoption graph (% of users on release)
  • 149.5.6 First-seen / last-seen per release per issue
  • 149.5.7 Regression-introduced-by release attribution
  • 149.5.8 Commit blame inside release diff
  • 149.5.9 Release comparison (A vs B delta)
  • 149.5.10 Automatic release creation from CI (Phase 145)
  • 149.5.11 Session tracking SDK (start / heartbeat / crash / exit)
  • 149.5.12 Session boundary rules (on background / after 30s idle)
  • 149.5.13 Release threshold alert (crash-rate > X %)
  • 149.5.14 Halt-deploy hook when threshold exceeded
  • 149.5.15 Rollback trigger via CI (Phase 145)

149.6 @neith/crash-sessions — Session Replay & User Context#

  • 149.6.1 DOM recording (privacy-safe masked input)
  • 149.6.2 CSS + asset snapshot for faithful replay
  • 149.6.3 Console + network + click timeline
  • 149.6.4 Mobile screen recording (native view tree)
  • 149.6.5 Redaction rules (mask / block / custom selector)
  • 149.6.6 Replay rate-limit / sampling per session
  • 149.6.7 Replay player with scrubber + speed
  • 149.6.8 Replay tied to specific event / error ID
  • 149.6.9 Rage-click / dead-click / error-click detection
  • 149.6.10 Accessibility tree overlay in replay
  • 149.6.11 Network waterfall inside replay
  • 149.6.12 User feedback widget with screenshot
  • 149.6.13 Cookie / localStorage snapshot at error
  • 149.6.14 Replay retention by plan
  • 149.6.15 Replay export as MP4 for ticket attachment

149.7 @neith/crash-performance — Tracing & APM Bridge#

  • 149.7.1 OTel-compatible distributed tracing ingest
  • 149.7.2 Span / transaction view with flamegraph
  • 149.7.3 N+1 query detection
  • 149.7.4 Slow-DB / slow-HTTP span classification
  • 149.7.5 Latency percentile (p50 / p75 / p95 / p99)
  • 149.7.6 Frontend Web Vitals (LCP / FID / CLS / INP / TTFB)
  • 149.7.7 Mobile startup trace (cold / warm / resume)
  • 149.7.8 Profile aggregation (sampled CPU profiles)
  • 149.7.9 Correlate trace → error → replay in one view
  • 149.7.10 Anomaly detection on latency regression

149.8 @neith/crash-alerting — Alerts, Notifications & Integrations#

  • 149.8.1 Rule builder (issue / metric / frequency / threshold)
  • 149.8.2 Escalation policy integration (Phase 152)
  • 149.8.3 Digest summaries (daily / weekly)
  • 149.8.4 Slack / sovereign chat (Phase 139) destination
  • 149.8.5 Email / SMS / push destination
  • 149.8.6 Webhook destination with HMAC signing
  • 149.8.7 Issue-owner routing (by path / tag / team)
  • 149.8.8 Auto-assign from commit author / CODEOWNERS
  • 149.8.9 Noise suppression (spike protection)
  • 149.8.10 Alert cooldown + silence windows
  • 149.8.11 Jira / Linear / sovereign PM ticket create
  • 149.8.12 PR comment on related commits
  • 149.8.13 GitHub / sovereign forge (Phase 12) issue link
  • 149.8.14 Teams / Discord / Mattermost destination
  • 149.8.15 Mobile app for on-call with ack / assign

149.9 @neith/crash-security — PII, Compliance & Self-Host#

  • 149.9.1 Server-side PII scrub (regex + named rules)
  • 149.9.2 Data scrubbing preview (see masked)
  • 149.9.3 Data residency per project (region pin)
  • 149.9.4 SOC 2 / HIPAA / GDPR audit controls
  • 149.9.5 Customer-managed keys (CMEK) encryption at rest
  • 149.9.6 Retention enforcement jobs
  • 149.9.7 Right-to-erasure GDPR API
  • 149.9.8 Consent-aware SDK (pause collection on opt-out)
  • 149.9.9 Self-host deployment profile (single-node / HA / regional)
  • 149.9.10 Air-gapped deployment option

149.10 @neith/crash-intelligence — AI-Assisted Triage#

  • 149.10.1 LLM suggest-fix (Nous Phase 47) on stack trace
  • 149.10.2 Similar-issue clustering across tenants (opt-in)
  • 149.10.3 Auto-PR proposed fix for known patterns
  • 149.10.4 Root-cause narrative (LLM summary)
  • 149.10.5 Severity inference from user-impact + trace
  • 149.10.6 Known-noise classifier (auto-ignore / auto-resolve)
  • 149.10.7 Flaky-test vs real-bug discriminator
  • 149.10.8 Embedding search ("find issues like this one")
  • 149.10.9 Feedback loop fine-tunes triage model
  • 149.10.10 Suggested owner via code-graph + history

149.11 @neith/crash-ui — Dashboard & Reporting#

  • 149.11.1 Issue feed with filters + saved views
  • 149.11.2 Issue detail (frames / context / breadcrumbs / replay)
  • 149.11.3 Trend graphs (events / users / sessions)
  • 149.11.4 Cross-project rollup dashboards
  • 149.11.5 Team-ownership dashboards
  • 149.11.6 Executive weekly report PDF
  • 149.11.7 SLO-style burndown of unresolved issues
  • 149.11.8 Command palette for power users
  • 149.11.9 Keyboard-first navigation
  • 149.11.10 Dark / high-contrast / reduced-motion a11y

149.12 Testing & Validation#

  • 149.12.1 SDK integration test matrix (15 platforms)
  • 149.12.2 Symbolication fidelity vs Sentry reference ≥ 99 %
  • 149.12.3 Ingest throughput 10M events / day soak
  • 149.12.4 Grouping accuracy vs labelled corpus ≥ 95 %
  • 149.12.5 Replay size / fidelity benchmarks
  • 149.12.6 PII scrub regression tests
  • 149.12.7 Drop-in Sentry DSN compatibility conformance
  • 149.12.8 Dogfood: serve every Oshun service (Phase 3–147) blocked:upstream Tagged 2026-09-18 against the range the item names, counted rather than assumed: 15 of the phase trackers between 3 and 147 still hold open items, TODOS/phase-27.md alone 934. "Every Oshun service" is not a set that exists yet, and a dogfood over the subset that does would be a different claim from the one this box makes. The seven sibling gates in §149.12 — symbolication fidelity, ingest soak, grouping accuracy, replay benchmarks, PII scrub, DSN compatibility — are the ones measurable here, and all seven are checked.