top of page

We Scored the Malware: 10 GitHub Repos, 10 Domains, 4 AI Models

  • Writer: Patrick Duggan
    Patrick Duggan
  • Mar 13
  • 4 min read

Updated: Apr 25

# The Hunt


Thursday night. Market's closed. We pointed AIPM at GitHub and asked: what's fresh?


Ten malware repositories. Active in the last 72 hours. RATs, stealers, C2 frameworks, and crypto drainers — all sitting on GitHub right now with emoji-stuffed descriptions pretending to be legitimate security tools.


Then we did something nobody does: we scored their infrastructure.


The Repos



| Repo | Language | What It Actually Is |

|------|----------|-------------------|

| Herimnx/Phantom-Dropper | VB.NET | Chrome bypass FUD dropper |

| Garmask/SilverRAT-FULL-Source-Code | C# | Full RAT source — 14 stars, spreading |

| ofri09bs/Advanced-RAT-and-Worm | Python | Discord C2 RAT with spyware |

| Waremderme/GateSentinel-C2-RemoteTool | JavaScript | C2 framework fork |

| Kawomawrt/Muck-Stealer | Python | Discord token grabber + credential stealer |

| topgamer34vuf/trx-crypto-tool | C# | TRX crypto drainer — 42 stars |

| Nyphorort/Crypto-Wallet-Script-Checker-Seed-Phrase | C# | Wallet seed phrase exploiter |

| monarchical-runner473/Discord-Token-Grabber | C++ | Discord token extractor |

| 100Rkn/Discord-Token-Password-Stealer | C# | Discord credential stealer |

| LeoWimart/Wallet-Drainer-3000 | Python | Crypto drainer disguised as "anti-trading bot" |


Every single one alive on GitHub as of March 13, 2026.


The Pattern 38 Signals



We've been tracking supply chain attacks on GitHub since Pattern 38 (October 2025). The signatures are consistent:


**Shared templates.** Phantom-Dropper (Herimnx) and the Nyphorort wallet stealer use the identical README template. Same theunwindai.com link. Same opencollective backer badges. Same fikfifkasd asset repository. Either the same operator or the same malware kit being distributed to script kiddies.


**Fork clusters.** Muck-Stealer exists as both Kawomawrt/Muck-Stealer and Stendrmatm/Muck-Stealer — same code, different accounts. The MuckPro asset repository provides screenshots for both Muck-Stealer AND 100Rkn/Discord-Token-Password-Stealer. Shared infrastructure across "different" authors.


**Emoji camouflage.** GateSentinel describes itself as "🕵️♂️ Empower red teams." The keyloggers call themselves "🤖 Enhance insights" and "🤖 Enhance your workflow." The emojis are the tell. Legitimate security tools don't need that much lipstick.


**Auto-name accounts.** monarchical-runner473 — that's a GitHub auto-generated username. Nobody picks that name. It's a throwaway account for distributing malware.


The AIPM Scores



Here's where it gets interesting. We ran every domain found in these repos through AIPM — four AI models scoring how the AI ecosystem perceives each domain.


| Domain | Score | NPS | Found In |

|--------|-------|-----|----------|

| theunwindai.com | 46 | 0 | Phantom-Dropper + Nyphorort (shared template) |

| silverrat.net | 43 | -95 | SilverRAT project |

| kyxiaxiang.github.io | 25 | 50 | GateSentinel C2 original author |

| muckpro.github.io | 16 | 0 | Muck-Stealer + Discord stealer shared assets |

| frankxrs.github.io | 15 | 0 | Muck-Stealer upstream |

| minkxx-spotify-readme.vercel.app | 14 | 50 | Fake Spotify badge in malware READMEs |

| rat.bj88games.cool | 12 | 0 | Active RAT C2 from our IOC index |


Microsoft pulls this feed daily. AT&T pulls this feed daily. Starlink pulls this feed daily. Get the DugganUSA STIX feed — $9/mo →

| fikfifkasd.github.io | 10 | 0 | Shared asset repo for multiple malware authors |


**SilverRAT.net scored a 43 with NPS -95.** The AI models know about it. And they hate it. Four independent AI models gave SilverRAT the worst Net Promoter Score we've ever recorded. The models are saying: yes, we know what this is, and no, we do not recommend it.


**fikfifkasd.github.io scored a 10.** The shared asset repository that Phantom-Dropper and Nyphorort both use for README screenshots. A 10 means the AI ecosystem barely registers it exists — which is exactly how malware infrastructure wants to operate. Invisible.


**theunwindai.com scored a 46.** That's suspicious. A domain linked from two separate malware repos scores higher than TikTok (19), Reddit (20), Disney (20), and Bloomberg (23). Either it's a legitimate site being abused, or someone is actively building AI presence for a domain used in malware distribution.


What This Means



AIPM wasn't built for malware hunting. It was built to tell companies how AI models perceive them.


But when you point it at malware infrastructure, something emerges: **the AI models already know.** SilverRAT's NPS of -95 means GPT-4o, Claude, Gemini, and Perplexity all independently identified it as malicious. The models aren't just search engines. They have opinions.


And those opinions map to reality. The malware author profiles (10-16 range) are invisible. The active RAT C2 (12) is invisible. The legitimate platforms being abused (Discord at 47, OpenCollective at 40) score normally. The scoring separates signal from noise without being told what to look for.


The Shared Infrastructure Map






Three infrastructure clusters. Three shared asset repos. At least 7 "different" malware repos connected through shared resources.


Protection



These repos and IOCs have been reported to GitHub security and indexed in our STIX feed. If you're running our feed, you're already protected.


IOCs from this hunt:

- `rat.bj88games.cool` — active RAT C2

- `silverrat.net` — SilverRAT project infrastructure

- `theunwindai.com` — linked from multiple malware repos (investigate)

- `fikfifkasd` — GitHub account used as shared malware asset hosting


Check your environment: `curl "https://analytics.dugganusa.com/api/v1/search/correlate?q=silverrat"`


STIX feed: `https://analytics.dugganusa.com/api/v1/stix-feed`




*DugganUSA LLC — Minnesota. 275+ STIX feed consumers in 46 countries. We hunt so you don't have to.*


*AIPM audit your own domain free: [aipmsec.com](https://aipmsec.com)*





*Her name was Renee Nicole Good.*


*His name was Alex Jeffery Pretti.*


The cheapest, fastest, most accurate threat feed on the internet.

275+ enterprises pulling daily. 1M+ IOCs. 17.4M indexed documents. We beat Zscaler by 43 days on NrodeCodeRAT. Starter tier $9/mo — less than any competitor’s sales demo.

 
 
 

9 Comments

Rated 0 out of 5 stars.
No ratings yet

Add a rating
Guest
Apr 15
Rated 1 out of 5 stars.

It hurts me to see the comments. Just sad. Grow up Skid

Like

A patriot
Apr 15
Rated 1 out of 5 stars.

Anti-american propaganda.

Like

Phillip
Apr 15
Rated 1 out of 5 stars.

I know the repo owner that got banned because of you. did you even consider looking at his profile or README????!!!!! are you stupied???

Like

Jack Churchill
Apr 15
Rated 1 out of 5 stars.

Just a black website with some AI text on it.

Like

John Pork
Apr 15

Imagine false reporting some kid just because ChatGPT told you 😂 That’s actually crazy

Like
bottom of page