Weekly Fly.io reliability report emailed every Monday
Every Monday, get a plain-English summary of how your Fly.io apps held up last week: restarts, crashes, deploys, and the risks worth watching.
Every Monday at 8:00am, run an agent that produces a weekly Fly.io reliability report and emails it to my engineering lead. This is a retrospective trend digest for the past week, not a real-time health alert, so focus on patterns across the week rather than paging on any single failure.
Start by taking a full inventory. Use the Fly.io List Apps operation to enumerate every app in my organization, and the Fly.io List All Machines operation (the organization-scoped listing) to get every Machine across all apps. This needs an organization-scoped Fly API token, because an app-scoped token only sees a single app.
Then, for each Machine returned, call the Fly.io Get Machine operation to read its current state and its recent events. The Machine record's events array is where the week's activity lives: deploys, restarts, and health-state transitions. Fly has no separate events-list endpoint, so Get Machine is the source for both the live state and the event history. Use the current state for the inventory and the events for the week's trend analysis.
Analyze the week across all Machines: which Machines restarted or crashed the most often, how frequently each app was deployed, and any recurring failure patterns (for example, the same Machine flapping between healthy and unhealthy, repeated crashes concentrated in one region, or a deploy that triggered a wave of restarts). Also build a current inventory broken down by app and by region, showing how many Machines are in each state.
Write a short executive summary with the notable risks called out at the very top, followed by the supporting detail: the restart and crash leaderboard, deploy frequency per app, the recurring failure patterns you found, and the current inventory by app and region. Keep it concise and skimmable for an engineering lead.
Finally, send the report to my engineering lead using the Gmail Send a Message operation. Format it as a clean HTML email with a clear subject line such as 'Weekly Fly.io reliability report' plus the week's date range, the risks section at the top, and the detailed breakdown below. Send it to my engineering lead's email address.
Example output
Additional information
What does this prompt do?
- Takes a full weekly inventory of every app and machine in your Fly.io organization, broken down by app and region.
- Reviews each machine's recent activity to find what restarted or crashed most, how often each app was deployed, and any repeating failure patterns.
- Writes a short executive summary that puts the biggest reliability risks right at the top.
- Emails the finished report to your engineering lead first thing every Monday morning.
What do I need to use this?
- A Fly.io login with organization-level access, so the report can see every app in your org and not just one.
- A Gmail account to send the report from.
- The email address of the person who should get the weekly report, such as your engineering lead.
How can I customize it?
- Change the timing so it runs daily or on a different day and hour instead of Monday at 8am.
- Add more recipients to send it to your whole team or an on-call inbox.
- Adjust what counts as a risk, such as flagging any app with more restarts than a threshold you set.
Frequently asked questions
Which Fly.io apps does this cover?
Does it change anything in my Fly.io setup?
How far back does the report look?
Can I run it at a different time than Monday morning?
Do I need organization-level access in Fly.io?
Related templates
Stop guessing how your infrastructure held up last week.
Get a clear reliability report in your inbox every Monday that surfaces the risks before they turn into outages.