Create systematic troubleshooting guides for technical issues

This skill turns an agent into a troubleshooting-guide writer that helps users systematically diagnose and resolve technical issues. It structures a guide around a problem-centric format: symptoms with exact error text, a 2-5 minute quick-checks section for the most common cause, then diagnostic steps with an if-normal/if-abnormal branch pointing to a specific solution, each solution carrying its own conditions, steps, verification, and rollback instructions, followed by prevention advice and an escalation section naming who to contact and what information to provide.

It applies several diagnostic flow patterns: a layer-by-layer approach for network issues (physical cabling, then ping/ipconfig at the network layer, then port testing and log review at the application layer), a binary-elimination method for performance degradation (test with a minimal configuration, then add components back one at a time until performance drops), and mermaid-diagram decision trees for multi-branch problems like a database connection failure (ping the server, then check credentials, then check whether the service is running, then check connection limits). It also documents error-message handling with cause-probability breakdowns (for example, "Connection refused" attributed 70% to the service not running, 20% to a firewall, 10% to the wrong port) and log-analysis commands for tailing and pattern-matching web-server, system, and application logs.

Use this skill when writing a guide to help users self-diagnose and fix a technical problem - network connectivity, a service outage, a database connection failure, or a performance regression - especially where a clear quick-check can resolve most cases before a deeper diagnostic sequence is needed. It insists each guide have testable commands, time estimates, rollback procedures, and defined escalation paths, and calls for user testing by someone unfamiliar with the problem before publishing. It is a documentation skill, not a live diagnostic or monitoring tool - the decision trees and log commands are guidance content, not something the skill executes itself.

Input: the problem to document, its symptoms, and the target audience's skill level. Output: a complete troubleshooting guide with symptoms, quick checks, a diagnostic decision sequence, multiple ranked solutions with rollback steps, prevention advice, and an escalation section - plus, for complex multi-branch problems, a mermaid flowchart decision tree.

The worked examples cover Linux/Mac (systemctl status, journalctl) and Windows (Get-Service, Get-EventLog) service diagnostics, network tools (ping, ipconfig, telnet, netstat), nginx error logs, and log-analysis one-liners using grep, cut, sort, and sed.

Support engineers, SREs, and technical writers producing self-service troubleshooting content for a product or service, and teams tracking guide effectiveness over time via resolution rate, time-to-resolution, and escalation rate.