breakfixlearn.com

AI-03AI & Prompt Engineering

Prompts for IT & networking tasks

Use this when: you want AI help with real IT work: reading output, explaining configs, writing scripts, tickets and runbooks.

Claude, ChatGPT and Gemini/v1.0//1 page/CC BY-NC 4.0
AI-03 Prompts for IT & networking tasks cheat sheet. Every table on it is repeated as text below.
Tap the sheet for the full-size image. The text version below is screen-reader friendly.

Text version

Everything on the sheet, as plain HTML: handy for screen readers, copy-paste and search.

6 templates to copy

Swap the <tags> for your own masked data.

#TaskPrompt
1Read the outputYou are a senior network engineer. Here is <output> from show ip interface brief. List the 3 most likely causes, the read-only check for each, and what each result means.
2Explain a configExplain this config line by line for a CCNA student. Flag anything insecure and say why. <config>...</config>
3Write a scriptWrite a Python 3 Netmiko script that backs up running configs from <device list>. Read credentials from environment variables, add error handling and a dry-run mode, and comment each step.
4Ticket to user emailRewrite this ticket note as an email to a non-technical user: what happened, what we fixed, what they need to do. Under 120 words, friendly. <note>
5Notes to runbookTurn these notes into an SOP with purpose, prerequisites, numbered steps, how to verify, and rollback. Ask me about any gap. <notes>
6Review a rule setReview this ACL for shadowed lines, rules that are too broad, and a missing final deny with log. Reply as a table: line, issue, fix. <acl>

Mask before you paste

  • Secrets: show run | exclude secret|password
  • IP addresses: swap them for 192.0.2.x and 198.51.100.x
  • Names: use user1 and example.com
  • Keys and tokens: delete them, every time

Make the answer checkable

  • "Put every command in a code block."
  • "Only show commands until I confirm."
  • "List what you assumed about my setup."
  • "Link the vendor doc for each command."

Follow-ups that improve any answer

  • "What could go wrong if I run this?"
  • "Give me the rollback steps too."
  • "Explain it like I'm new to this."
  • "Compare both options in a table."
  • "Turn this into a checklist."
  • "What would prove you wrong?"

Use AI safely at work

Danger. Never run an AI-written script or config on production until you have read it and tested it in a lab.

Caution. Use only the AI tools your company approves. Public tools may keep what you paste.

Caution. AI can invent flags and commands. Check each one in the vendor docs.

Remember

  • Paste the real output, masked
  • Say the device, OS and version
  • You own the change, not the AI

Sources

Go deeper on the blog