error-tracker

name	error-tracker
description	Use when the app needs database-backed error tracking without an external SaaS. Covers installing ErrorTracker with automatic Phoenix/LiveView/Oban reporting, migrations, pruning, and a dashboard.
user-invocable	true

ErrorTracker

Adds ErrorTracker, an Elixir-native error tracking solution, to a Phoenix app. It stores exceptions in your existing database and serves a LiveView dashboard for reviewing, resolving, muting, and inspecting occurrences — no external service or hosted account required. Phoenix controllers, LiveViews, and Oban jobs report automatically.

When to use

• The app needs to capture and review production exceptions but you don't want Sentry/AppSignal or another hosted error service.
• You want stack traces, occurrences, breadcrumbs, and context stored alongside your own data in your existing repo.
• Don't reach for this if you already pipe errors to a hosted tracker — running both duplicates noise. For a no-dashboard setup, plain Logger may suffice.

Dependencies

• Requires: None.
• Recommended: admin (mounts the dashboard under /admin/errors and links it from the developer page), oban (adds job metadata to error context).

Install

mix saaskit.feature.install error_tracker

No install decisions. The mount point is chosen automatically:

• With the admin feature installed → dashboard at /admin/errors, linked from the admin developer page.
• Without admin → dashboard mounted beside the development LiveDashboard route.

What it generates

• Adds {:error_tracker, "~> 0.9.0"} to mix.exs deps.
• Config block in config/config.exs (see Configuration).
• Migration priv/repo/migrations/*_add_error_tracker.exs — delegates to ErrorTracker.Migration.up/0 to create the error tracker tables.
• lib/my_app_web/router.ex — injects use ErrorTracker.Web, :router and the error_tracker_dashboard "/errors" route at the appropriate mount point.

Configuration

Injected at the top of config/config.exs:

# config/config.exs
config :error_tracker,
  otp_app: :my_app,
  repo: MyApp.Repo,
  enabled: Mix.env() != :test,
  plugins: [ErrorTracker.Plugins.Pruner]

• enabled — on everywhere except :test. Reporting is off during tests so expected failures don't pollute the dashboard.
• repo — uses your existing MyApp.Repo for storage; no separate database.
• plugins: [ErrorTracker.Plugins.Pruner] — runs with defaults to clean up resolved errors over time. Tune retention by passing options to the plugin.

Tweaking for your app

Report an error manually from a rescue/catch block with extra context:

try do
  perform_risky_operation()
catch
  error ->
    ErrorTracker.report(error, __STACKTRACE__, %{operation: "risky_operation"})
end

Configure pruning retention instead of using the defaults:

# config/config.exs
config :error_tracker,
  otp_app: :my_app,
  repo: MyApp.Repo,
  enabled: Mix.env() != :test,
  plugins: [
    {ErrorTracker.Plugins.Pruner, max_age: :timer.hours(24 * 30)}
  ]

Restrict who can open the dashboard by guarding the route in a pipeline that checks the current user's role:

# lib/my_app_web/router.ex
scope "/admin" do
  pipe_through [:browser, :require_admin]
  error_tracker_dashboard "/errors"
end

After install

• mix ecto.migrate — create the error_tracker tables.
• Visit /admin/errors (or /dev/errors if no admin feature) to confirm the dashboard loads.
• Trigger a fake error in dev to verify reporting works end-to-end.
• Decide on a notification channel for new errors (Slack webhook, email digest, etc.).