It’s no secret AI has led to a lot of micro-apps, and enterprise apps creating their latest feature with AI in the name. Well, I’ve been experimenting with Claude Code and other tools to see how I can actually spend less time doing menial tasks. Here’s three tools I’ve actually built that are saving me time, and some behind-the-scenes on my workflow.
Wi-Fi Tool Comparison Matrix
Over the years, so many new apps have come to fruition that I had a hard time thinking about which ones fit my use cases in a more broad sense. Great comparisons exists just amongst surveys tools or only throughput testers, but I wanted a quick chart for everything that a Wi-Fi engineer may encounter; some tools survey but don’t do spectrum analysis. Other tools analyze your uplink, but don’t look at channel utilization.
That led to me and Claude chat one evening. My goal that night was to figure out how to embed a vibe coded micro-app easily into a website. After a few corrective prompts, I had something almost usable; mobile view still broke. Because of that I left it there and actually forgot about the page until recently a few folks found it on LinkedIn.
In any case, if you’d like to try it, the link is here; unfortunately, I probably won’t get around to fixing the mobile view so desktop is the way to go.
ChromeOS Wi-Fi Tool
Around the same time, another problem plagued my conscience: Chromebook troubleshooting. Long story short, I had a school customer where I needed to debug Chromebook roaming decisions in order to fix their Wi-Fi design. Good news is that ChromeOS gives a treasure trove of logging about Wi-Fi roaming.
However, after scrolling through a few thousand lines to find my conclusion, I thought “there has to be an easier way”. At first, I just used AI to analyze the log file itself, but its analysis of the timestamps and the correlated logs were inaccurate at times. I still had to go read the file myself
Example of the log format
Following the project, I decided I never wanted to do that again and started again vibe-coding. At first, I used Google AI Studio but it felt clunky for my use case. After generating a PRD with it, I went to Claude Code.
Starting out vibe-coding I suggest doing the following:
Ask AI to create a PRD (Product Requirements Document)
Ask it to conduct security audits
Start from the beginning using Git for version control
Use a tool such as VSCode to make your life easier
For the security audit, you can use something like:
Act as an expert security researcher conducting a thorough security audit of my codebase. Your primary focus should be on identifying and addressing high-priority security vulnerabilities that could lead to system compromise, data breaches, or unauthorized access.
Follow this structured approach:
ANALYSIS PHASE:
Review the entire codebase systematically
Focus on critical areas: authentication, data handling, API endpoints, environment variables
Document each security concern with specific file locations and line numbers
Prioritize issues based on potential impact and exploitation risk
PLANNING PHASE:
For each identified vulnerability:
Explain the exact nature of the security risk
Provide evidence of why it's a problem (e.g., potential attack vectors)
Outline specific steps needed to remediate the issue
Explain the security implications of the proposed changes
IMPLEMENTATION PHASE:
Only proceed with code modifications after completing analysis and planning
Make minimal necessary changes to address security issues
Document each change with before/after comparisons
Verify that changes don't introduce new vulnerabilities
Key Focus Areas:
Exposed credentials and environment variables
Insufficient input validation
Authentication/authorization bypasses
Insecure direct object references
Missing rate limiting
Inadequate error handling and logging
Unsafe data exposure
DO NOT:
Make cosmetic or performance-related changes
Modify code unrelated to security concerns
Proceed with changes without explaining the security implications
Skip the analysis and planning phases
After each modification, explain:
What security vulnerability was addressed
Why the original code was unsafe
How the new code prevents the security issue
What additional security measures should be considered
Read these articles on vibe coding security:
https://www.dryrun.security/resources/owasp-top-10-llm-building-secure-applications?__hstc=17958374.cb2733a3cf252bf262754b3b399c8151.1764885622593.1771875517351.1771881334807.43&__hssc=17958374.9.1771881334807&__hsfp=fee2e0da729724e2135f129d69d46967
https://docs.replit.com/tutorials/vibe-code-security-checklist
Use the articles and these security issues with their best practices to evaluate the code-base and remediate security issues:
Trusting Client Data: Using form/URL input directly.
Fix: Always validate & sanitize on server; escape output.
Secrets in Frontend: API keys/creds in React/Next.js client code.
Fix: Keep secrets server-side only (env vars, ensure .env is in .gitignore).
Weak Authorization: Only checking if logged in, not if allowed to do/see something.
Fix: Server must verify permissions for every action & resource.
Leaky Errors: Showing detailed stack traces/DB errors to users.
Fix: Generic error messages for users; detailed logs for devs.
No Ownership Checks (IDOR): Letting user X access/edit user Y's data via predictable IDs.
Fix: Server must confirm current user owns/can access the specific resource ID.
Ignoring DB-Level Security: Bypassing database features like RLS for fine-grained access.
Fix: Define data access rules directly in your database (e.g., RLS).
Unprotected APIs & Sensitive Data: Missing rate limits; sensitive data unencrypted.
Fix: Rate limit APIs (middleware); encrypt sensitive data at rest; always use HTTPS.This process spanned 2-3 weeks as I kept burning through my token limits. Once I had something I wanted to share, I made a GitHub page here. The latest version adds a chart for each AP roam. Go give a try and let me know what you think.
Connection Timeline
Auto-generating Topologies With Draw.io
This last use case realistically saves me the most time. There’s nothing that drives me crazier than struggling in diagraming tools to get lines and boxes to the exact pixel needed x100 for larger networks.
Enter draw.io (Link)
It’s a free diagramming tool that has been around a long time. Every draw.io file uses XML and because of that, I thought maybe there’s some way to automate this.
Full disclosure: I did this at work using a tool called Glean. This matters for two reasons:
Glean uses an AI instance with internal company knowledge
The agents feature allows for selection Claude Sonnet, ChatGPT, Gemini etc.
Most importantly, I don’t hit token limits coding this way
Due to the above, this process is a lot faster to get accurate topologies, but I also can’t share my results yet.
Here’s how you could make this tool this for any deployment (and how I might in the future):
Program flow for automated topologies
From there, you could add extra features like a cabling legend or port labels. I’m still trying to get my prototype working so this is a WIP.
So that’s three ways I’ve recently used AI to build tools that actually save me time. I hope this inspires y’all out there to experiment. Until then, I have more tokens to burn.
Until next week,
Eva
Need help starting your Wi-Fi career?
My new Wi-Fi interview prep kit is in my store here. Get the career bundle (career cheat sheet + prep kit) for 20% off full price.