For app owners
How DeployAngel handles your app
The team that builds and maintains your app uses DeployAngel to check that each new release works in production, and that its scheduled jobs keep running. This page explains, in plain terms, what that involves for your app and its data.
What it does
A small agent runs inside your app. It counts what the app does, such as requests, response times, errors, and background jobs, and sends a summary to DeployAngel about once a minute. When a new release goes out, DeployAngel compares it with how the app ran before and tells your team whether it's healthy. It also notices when a scheduled job, like a nightly invoice run, stops running.
What it sends
- Counts, status codes, and response times for each kind of page or API call, by its pattern, such as
GET /orders/:id. IDs in addresses are never sent. - Counts, failures, and durations for each kind of background job.
- Errors: the type of error, where in the code it happened, and the first line of its message, with numbers, IDs, email addresses, and quoted text replaced by placeholders. Other words are kept, so a message the app builds with a customer's name in it would include that name. Your team can turn messages off entirely.
- Which release is running, and which server sent the summary.
- A description of the app's code: its routes, job names, scheduled jobs, framework versions, and file names with fingerprints of their contents, so DeployAngel can tell what changed between releases. Never the code itself.
What it never sends
Request or response contents, form fields, query parameters, headers, cookies, session data, passwords, database records, or SQL values. DeployAngel is built not to receive personal data about your app's users.
How it affects your app
- Speed. Recording a request adds about 2 microseconds. Summaries are sent from a background thread, so no request waits on the network.
- Reliability. If DeployAngel is slow or down, your app carries on as normal. At worst, some summaries are lost.
- No way in. DeployAngel never connects to your app, sends it traffic, or stores credentials for it. It can't deploy, roll back, or change anything in your app or your servers.
- Open source. The agents' code is public on GitHub, so anyone can check what they send.
Where data is kept, and for how long
- DeployAngel runs on DigitalOcean. All traffic to and from it uses HTTPS.
- Detailed telemetry is kept for 21 days. Release history and daily totals are kept for 7, 30, or 90 days, depending on your team's plan.
- If your team removes your app from DeployAngel, its data is deleted right away, and from the encrypted database backups within 7 days.
- Each team's data is kept separate, and only its members can see it. Secrets, like the addresses alerts are sent to, are encrypted.
AI
AI investigation is optional and off by default. If your team turns it on, then when a release fails, a small bundle of evidence (the failed checks, new errors, and what changed in the release) goes to Anthropic, which writes a likely explanation. Under Anthropic's commercial terms, that data isn't used to train its models. AI never decides whether a release passed.
Who runs it
DeployAngel is an independent service run by Jordan Owens in Ohio, USA. It's in beta. The privacy policy lists everything it collects, every provider that processes data, and your rights, and it's the binding version of everything on this page. Questions go to DeployAngel's contact form, or to the team that maintains your app.