Features
Every release gets a verdict, decided by fixed rules from real production traffic. All of this is included, and free while DeployAngel is in beta.
A verdict for every release
- ✓Cleared, Failed, or Not cleared yet, decided by fixed rules you can read, never by AI
- ✓Compared with your last healthy release, release by release
- ✓A clearance report: what ran, what was compared, what couldn't be observed, and what's still being watched
- ✓Confidence and coverage on every verdict
- ✓Clearance revoked if a watched job misses its run later
- ✓Deploying faster than releases clear? The next verdict covers everything since the last cleared release, and the dashboard says when nothing has cleared for a while
- ✓What changed, from GitHub: pull requests, dependency and framework upgrades, migrations, and the files most likely behind a failure. Optional and read-only
- ✓AI investigation: a plain-language explanation of why a release failed, from the evidence. Off until you turn it on, and it never decides a verdict
What it checks
- ✓HTTP 5xx rate and p95 latency, app-wide and per route
- ✓Job failures, duration, and queue latency for Solid Queue, Sidekiq, and any ActiveJob adapter
- ✓New exceptions, and spikes in existing ones
- ✓Checkpoints for business events that should keep happening
- ✓Critical flows you name, like sign-in or checkout
- ✓Rarely used paths, with an alert if they fail on first use
- ✓Your own smoke tests, reported as external checks
Fits how you ship
- ✓Alerts by email, Slack, or signed webhooks
- ✓A Monday email: each app's releases that week, and the rarely used paths no clearance has covered yet
- ✓CLI and MCP server for CI and coding agents, with exit codes a pipeline can act on
- ✓Staging environments, with quiet alerts by default
- ✓Kamal, Docker, Render, Fly.io, Railway, Coolify, Dokku, Capistrano, and Heroku
- ✓Unlimited apps and teammates