Signing in
| Symptom | Fix |
|---|
| "Confirm your email" keeps appearing | Enter the 6-digit code, or request a new one. Check spam |
| Password does not work | Use Forgot password. Team members must ask their admin to reset it |
| Signed out unexpectedly | Someone changed the password on the account, or the 30-day session expired |
| Team member cannot sign in | They must use the Team member tab with the Account ID, not the Personal tab |
| Team member is refused everywhere | They have not completed the forced password change |
Servers
| Symptom | Fix |
|---|
| Cannot add a server | Connection problems covers every error |
| Was working, now "Can't connect" | IP changed, password changed, key removed, firewall changed, or disk full |
| Session drops after a while | Normal: 30 idle minutes. Click back in |
| Everything disconnected at once | The app restarted. Reopen |
| A button fails with "permission denied" | The SSH login you configured lacks the rights. Not a DevOps Agent problem |
| A tab is missing or buttons are absent | Your role. See Roles |
| Health card shows nothing | The first check is still running, or the server is unreachable |
The assistant
| Symptom | Fix |
|---|
| "Needs Execute mode" | It wants to change something. Switch the mode if you agree |
| It will not touch a server | The server is not attached to this chat, or it is in another project |
| Out of credit | Use your own key, or ask your admin |
| Answers are vague | Name the server, say what you see, use a guided task |
| It ignores an integration | Mention the service by name — the tools it loads depend on what you ask about |
| Stop does not stop it | A known issue: the browser stops listening but the command may finish. Check the activity log for what actually ran |
| An approval card vanished | The app restarted. Nothing ran. Ask again |
| A long job is cut off | Bot commands are stopped after 20 minutes |
Uptime
| Symptom | Fix |
|---|
| Says DOWN, site looks fine | Cached copy, intermittent failure, or a wrong expected status. See Reading the results |
| No certificate information | The site is HTTP, or TLS is terminated by a CDN |
| No alert when the site went down | There are no alerts yet. See Uptime |
AWS
| Symptom | Fix |
|---|
| Nothing loads | Key or secret wrong. Replace in Settings → Project |
| One region errors | Missing permission, or the region is not enabled |
| Cost charts empty | Enable Cost Explorer once in AWS, and add ce:GetCostAndUsage |
| "not authorized to perform" | The exact action is named. See Permissions |
| EC2 list is stale | Press Refresh — the list is cached |
| Cannot SSH to an EC2 instance | Security group must allow port 22, and the instance needs a public address |
Integrations
| Symptom | Fix |
|---|
| 401 / 403 | Wrong or expired token; or the account lacks access |
| Timeout | The service is not reachable from the internet |
| Grafana panel blank | Use a public dashboard; check allow_embedding |
| MCP tools do not appear | Re-save to re-test the connection; check the token |
Still stuck
Email support@devops-agent.io with the screen you were on, what you
expected, what happened, and the exact message. If it involves the assistant,
say which chat — the full record of what it ran is stored and we can read it.
