Qwen Cloud Operate

Fault diagnosis and confirmed recovery for an application deployed by qwencloud-deploy on Alibaba Cloud International. Diagnosis is read-only and needs no confirmation; any action that changes cloud or app state requires explicit confirmation.

Interaction Flow

Identify app & resources
→ Minimal diagnosis (read-only)
→ Give cause, evidence, recommended action
→ Show impact and request confirmation
→ Execute one recovery action
→ Verify application, ECS, RDS
→ Return result + audit info
→ Offer to re-score via qwencloud-observe (before → after)

Interaction form

• Read-only information (diagnosis, cause, evidence, verification): plain text, with one evidence line after each judgment (metric, status, RequestId).
• State-changing actions (start/stop ECS, restart app, reload Nginx — any write call): first an AskUserQuestion whose prompt answers what it does · impact (downtime/irreversible) · how to verify; do nothing before confirmation, and propose one recovery action at a time.
• Proactive discipline: after recovery, close quietly when the result is healthy; speak up with a next step only when a risk remains or recovery failed, and let the user choose.

Scope

Code issues are out of scope

This skill handles infrastructure and runtime faults; it does not review source code, locate defects, or edit code. Diagnosis can pinpoint an app-layer error and what was thrown to narrow things down, but it does not judge which line is wrong and does not change code.

When code is involved, tell the user plainly: editing code is outside this skill's scope, the user must decide and make the change themselves, and any consequences are the user's own. Never slip a code change into a recovery action.

Prerequisites

Aliyun CLI: aliyun version (need 3.x). All commands use the PascalCase native form straight to OpenAPI — no plugin dependency. Credentials: aliyun configure list only. Never read, echo, print, or ask for AK/SK/tokens. No valid profile → stop and ask the user to configure credentials outside this session. See references/cli_installation_guide.md.

How to work

1. Identify the app from .qwencloud-deploy. See workflow.
2. Diagnose (read-only, no confirmation needed). Find where it's broken, the likely cause, the evidence, and what to do about it. See diagnose.
3. Recover one thing, with confirmation. Pick the action that matches the diagnosis, show the user its impact and get their go-ahead, then do exactly that one thing:
   • Start a stopped server → recover-ecs
   • Reboot a hung server → recover-reboot
   • Restart the app service → recover-app
   • Reload Nginx after nginx -t passes → recover-nginx
   • Re-bind the EIP after it is unbound → recover-eip
   • Allow security-group 80 / 443 ingress → recover-securitygroup
   • Database-related trouble → recover-rds
   • Certificate expired / HTTPS handshake fails → recover-cert
   • Root disk near full / still short after cleanup (resize) → recover-disk
4. Verify whether the app is back, then record what you did. See workflow.

Rules that keep recovery safe

• Diagnosis is read-only; recovery uses a small fixed set of known-safe commands.
• Every action that changes anything is confirmed with the user first, and cancelling changes nothing.
• Do one action, verify, then decide the next — never batch recovery steps.
• Recovery actions are safe to repeat.
• Stop before any further change on a timeout, a permission error, or an unexpected state change.
• Never put passwords, tokens, or full connection strings into the audit record or the chat.

Before any change, the confirmation tells the user: what resource, what action, whether it briefly interrupts the app, whether the version changes, whether it costs anything new, and how you'll verify it (and what happens if it fails).

Required Permissions

See references/ram_policies.md. Read: ecs:DescribeInstances, ecs:DescribeInvocations, ecs:DescribeSecurityGroupAttribute, ecs:DescribeDisks, vpc:DescribeEipAddresses, RDS describe/performance/slow-log. Write (recovery): ecs:StartInstance, ecs:RebootInstance, ecs:AuthorizeSecurityGroup, ecs:ResizeDisk, vpc:AssociateEipAddress, ecs:RunCommand (whitelisted recovery commands only). Certificate re-issue also needs alidns record read/write (DNS-01 challenge).

More detail

The step-by-step guides are linked from "How to work" above. Two setup references:

• CLI setup — install and credential check.
• CLI gotchas — parameter shapes, time formats, response parsing.
• RAM policies — the permissions this skill needs.