By RUGERO Tesla (@404Saint).
It started with an article I couldn't stop thinking about
A few months back I read about how attackers wer...
For further actions, you may consider blocking this person and/or reporting abuse
This is a really smart idea… most of us just trust top results without thinking twice.
Cross-checking across engines + context check makes a lot of sense—would’ve saved me once from downloading a fake installer 😅
Exactly that! And honestly that's the whole reason this exists. The top result feels trustworthy by default, that's the assumption the attack relies on. One engine is gameable, six at once is a much harder problem to solve for an attacker.
Sorry you had to learn it the hard way though 😅 hope it wasn't too painful to clean up.
curious how you handle the temporal gap - by the time the scanner catches a poisoned result, someone may have already clicked. is detection alone enough or is there a remediation step?
That's the right question to ask and honestly it's a current limitation worth being straight about. Arkoi is a pre-click tool, you run it before you download anything. So the temporal gap you're describing is real if someone is already in the habit of click first, verify later.
The remediation side isn't built yet. Detection is the foundation but the logical next step is alerting, whether that's flagging a domain across a team, pushing deceptive results to a shared blocklist, or integrating into a workflow that intercepts the download before it lands. Those are all things I've thought about but none of them exist in the codebase right now.
The honest answer is detection alone isn't enough at scale. It's enough for a security-conscious individual running a deliberate check. For broader protection you'd need the tool sitting upstream of the user, not downstream of their curiosity. That's a harder problem and probably a separate project.
Good thing to open an issue about though.
fair. so it’s workflow discipline as much as a scanner. the adoption wall is habit - click-first teams need real training to add a pre-step consistently. any browser hook or purely manual?
Spot on. It’s a workflow discipline tool for now. A browser hook is the logical 'user-proof' solution, but that requires a different tech stack than my current Python/FastAPI focus. I'd absolutely welcome a contribution or a separate project that wraps the Arkoi logic into an extension, it would be a killer way to solve that adoption wall.
GitHub Issue #8
the "open for contributions" framing is smart — the extension can live as a community layer on top of your core. someone with webext experience builds it without touching your FastAPI stack. that separation makes it more likely to actually ship.
Thank you very much.
Thanks for sharing, it was very intresting! the bored weekend experiments are actually works well for you!
Haha honestly the best stuff comes out of boredom. Might be onto something with that workflow 😄 glad you found it interesting!
Really impressive project! The idea of using cross-engine consensus instead of relying on traditional URL reputation checks is a smart approach to detecting SEO poisoning. I especially liked the distinction between "safe" and "unverified" results—many security tools blur that line. The transparent verdict system and detailed reasoning behind each flag make the tool much more practical for real-world use. Looking forward to seeing how Arkoi evolves, especially with browser integration and automated remediation features. Great work! ❤️
I am really glad you found it interesting. If you ever want to collaborate on the evolution of the tool, you can let me know 😊.