Incident playbooks
Playbooks are how mttrly acts on a server: predefined, scoped operations in the agent — more than 80 built in — combined into diagnostic recipes for common incidents. Runtime policy, not the source folder name, determines whether an operation can run read-only or must stop for approval. The guides below take frequent incidents and show both paths: how to fix each one manually over SSH, and how mttrly runs the same triage from chat or an AI IDE. Approval-required fixes normally stop for human confirmation. A separately user-authorized Investigation can skip per-command confirmation only for mttrly_execute_command within its server, time, and action-count bounds; configured preauthorization applies only to explicitly enabled automation.
Linux server disk is full
What to check when df shows 100%, which cleanup targets are safe, and how the disk-full recipe finds them — with cleanup behind an explicit approval.
Website is down
A fixed diagnosis order — ports, nginx config test, SSL, logs — and how the website-down recipe runs it in one pass, with restarts behind approval.
Nginx 502 Bad Gateway
Separate nginx from the upstream failure, inspect service and container evidence, then request only the bounded restart or reload the diagnosis supports.
Docker container keeps exiting
Inspect exit state, logs, memory pressure, and image reality before requesting an approval-gated container restart.
Production broke after deploy
Use the code-backed post-deploy-issue recipe to correlate recent logs, health, and memory before choosing a forward fix or rollback.
Roll back a deployment from your phone
Review the current release, request the managed rollback path, approve it from mobile, and verify service health afterward.
Detection tools tell you something is wrong. mttrly is the agent on your server that diagnoses the incident and prepares the fix. Read-only diagnostics run directly; approval-required changes normally create a pending action. The explicit exceptions are a user-authorized, bounded mttrly_execute_command Investigation and narrowly configured preauthorization, and both remain audited. See how mttrly compares to monitoring tools or how teams manage a VPS without raw SSH, connect the incident response action layer, or add Telegram for mobile approvals.
Put the playbooks on your server
Connect a server in a few minutes. Free monitoring and read-only diagnostics to start; approval-gated remediation when you want it.