denyzhirkov/crashbox repository overview

⁠crashbox

Crashbox accepts events from unmodified official Sentry SDKs (@sentry/node, @sentry/browser, …) by changing only the DSN. No custom SDK, no Postgres+Redis+Kafka+ClickHouse stack — just one container.

Not a full Sentry replacement: no performance traces, session replay, source maps, attachments, orgs/teams. If you need any of those, look elsewhere. If you want an error inbox for your side project tonight — this is it.

Source on GitHub⁠ · Docs⁠

⁠Quick start

docker run -d --name crashbox \
  -p 8080:8080 \
  -v crashbox-data:/data \
  -e [email protected] \
  -e CRASHBOX_ADMIN_PASSWORD=change-me \
  -e CRASHBOX_PROJECT_NAME=my-app \
  -e CRASHBOX_PUBLIC_URL=http://localhost:8080 \
  denyzhirkov/crashbox:1.1.0

Watch the logs once for the DSN:

INFO bootstrap: project DSN (shown once at startup)
     dsn=http://<key>@localhost:8080/1

Point any Sentry SDK at it — only the DSN changes:

import * as Sentry from "@sentry/node"
Sentry.init({ dsn: "http://<key>@localhost:8080/1" })

Open http://localhost:8080, log in, your crashes are there.

⁠Tags

⁠Required environment

⁠Useful optional environment

Full reference: docs/configuration.md⁠.

⁠What you get

• Ingest: POST /api/:project_id/envelope[/] compatible with @sentry/node, @sentry/browser, and any SDK using the standard DSN/envelope format
• Web UI: Login → Projects → Issues list (with filters, search, tag click-to-filter, saved views) → Issue detail (stack trace + breadcrumbs + tags + user + request + raw JSON). Warm-dark theme. Cmd+K command palette anywhere. 24h sparkline per issue.
• Issue management: resolve / reopen / snooze (1h / 1d / 1w / until-next-crash) / auto-resolve / auto-reopen
• Alerts: webhooks for new issue / re-open / spike detection (rate jumps 5× over 24h baseline) — Telegram, Discord, or any URL
• Operations: built-in CLI (docker exec crashbox crashbox projects list, issues resolve <id>, backup /data/snap.db), Prometheus /metrics, atomic SQLite backups via VACUUM INTO
• Single container: 40 MB distroless image, non-root, statically-linked SQLite, no shell

⁠Volume

/data — SQLite database, WAL file. Mount a named volume or a host path so you don't lose data on container restart.

⁠Health

GET /healthz — liveness · GET /readyz — readiness · GET /metrics — Prometheus

# minimal compose
services:
  crashbox:
    image: denyzhirkov/crashbox:1.1.0
    restart: unless-stopped
    ports: ["8080:8080"]
    volumes: [crashbox-data:/data]
    environment:
      CRASHBOX_PUBLIC_URL: "http://localhost:8080"
      CRASHBOX_ADMIN_EMAIL: "[email protected]"
      CRASHBOX_ADMIN_PASSWORD: "change-me"
      CRASHBOX_PROJECT_NAME: "my-app"

volumes:
  crashbox-data:

⁠Image details

• Base: gcr.io/distroless/cc-debian12:nonroot — glibc + ca-certificates + a non-root user, no shell
• Size: ~40 MB
• User: nonroot:nonroot (uid 65532)
• Entry: /usr/local/bin/crashbox (same binary, serve is default; runs as CLI for other subcommands)
• Architectures: linux/amd64 (multi-arch coming later)

⁠License

MIT or Apache-2.0 at your option.