Iam 12 .I Built an Autonomous Bot to Hunt Phishing Scams on Dev.to (For $0) 🐯🛡️

Hey DEV Community! 👋

Recently, I noticed a nasty phishing bot (namespamming the comment sections of popular posts with fake "Official Platform Update" links. It was targeting new developers and trying to steal credentials.

I started manually flagging and replying to warn the community. But as a 12-year-old founder building a 142KB AI app on a $150 phone, I realized two things:

1. Manual reporting doesn't scale. The bot posts faster than I can click.
2. Running a background script on my phone would drain my battery and data cap.

So, I did what any constraint-driven engineer would do: I built an autonomous, cloud-native security bot to hunt it down for me.

Meet Tiger Sentinel (running as @watch_dog). 🤖

️ The Architecture (Zero Cost, 100% Autonomous)

I didn't want to pay for a VPS, and I didn't want to keep my phone awake. The solution? GitHub Actions.

Here is the exact stack I used to build a 24/7 autonomous guardian for $0:

1. The Burner Account (OpSec): I created a dedicated Dev.to account (@watch_dog ) to hold the API key. If the bot accidentally triggers a spam filter, my main founder account (`@harun_solo_dev ') stays 100% safe.
2. The Brain (Python + Dev.to API): I wrote a lightweight Python script using the requests library. It hits the Dev.to API, fetches the 3 most recent articles, scans the comments for the scam signature (tr.ee/dev-to), and posts a warning if found.
3. The Security (GitHub Secrets): I stored the Dev.to API key securely in GitHub Repository Secrets, so it's never exposed in the code.
4. The Trigger (Cron Job): I set up a GitHub Actions workflow to run the script every 15 minutes.

🛡️ The "Anti-Spam" Logic

Building a bot that replies automatically is dangerous. If you reply too fast, the platform bans you. I had to engineer human-like constraints into the code:

• Duplicate Checking: Before replying, the bot checks if @watch_dog has already replied to that specific article. This prevents infinite spam loops.
• Rate Limiting: I added a time.sleep(random.randint(3, 7)) delay between fetching and posting to mimic human behavior and respect the API rate limits.
• Scope Limiting: It only scans the top 3 recent articles per cycle. It catches the scam in real-time without digging up old posts and spamming the feed.

💡 The "Tiger" Philosophy

Building Tiger Sentinel taught me a massive lesson about product development: Don't just build features; build solutions to real ecosystem problems.

When you build tools that protect, empower, or save time for your community, you aren't just writing code. You are building trust. And trust is the ultimate distribution channel.

🚀 What's Next?

While Tiger Sentinel guards the gates in the cloud, I am pivoting 100% of my energy into Pure Distribution for my main projects:

• KODA: A 142KB, mobile-first AI code mentor with live voice chat and study rooms. Built entirely on a $150 phone for users on spotty 3G. Try KODA Here
• Off Radar: A tiny, constraint-driven newsletter for devs building powerful tools on a budget. 👉 Subscribe to Off Radar

If you're building on a budget, or just love seeing how a 12-year-old automates their life with Python and GitHub Actions, let me know in the comments!

Stay sharp, verify your links, and keep building. 🐯️

Comments

[Amn Tree]

You're a seriously talented young developer! I'm genuinely impressed, not just by your vision, but especially by your determination to turn an idea into something real and actually useful.

Having ideas is one thing. Building something that works, solves a real problem, and helps other people is a completely different story.

And doing all that at 12 years old? That's impressive!

Keep that curiosity and determination alive. You're definitely on the right track.

Great job! Huge respect! 👏

[Harun - solo dev]

amn, thank you so much for the incredibly kind words! 🐯

honestly, the best way to learn is by building things that actually solve real problems. seeing that the tiger sentinel is actively protecting the community makes all the debugging completely worth it.

i'm just getting started, and i'm channeling all this energy into my main project, koda (a 142kb, zero-bloat ai code mentor built entirely for budget phones). if you ever want to check it out, roast my ux, or just chat about building on a constraint-driven budget, i'd absolutely love to connect.

thanks again for the massive encouragement. it means the world! 🚀