Don't just partner with someone you met on the internet right away. Do a project with them. Collaborate on some throw away stuff. Write a to do list together , whatever. The idea is to see if they can even commit to something like that. Build something with them even if not a software project. Go hang out with them, socialize, hit the bars whatever. See how they are.
I really think that a cofounder cannot be magically "found" but more like you worked with someone for a while and realized that they could be a good potential cofounder. You already trust them and know their work. Otherwise, you are taking a huge risk and it almost always ends up bad for both parties.
Yes, I am sure outliers are there but they are outliers. If you really want to find a co-founder, make sure you have enough time with them as non-cofounder before you even think of moving on with that step.
Remember, the real test of a "partner" in anything is not when things are great. It is when things go south and I can assure you they WILL. How your partner/cofounder behaves during the tough times will tell you all about them. You hope to have an idea about that before you even commit to be a cofounder.
You don't identify toxic people based off how they interact with you. You identify them from the impact craters they leave behind. (And if they figure out how hide by no longer causing damage, then good - that was the entire point).
For example take 10 developers, even if they are all great they will have different strong points. Some devs will have a strong product design/UX intuition, others will be strong at building platforms/performance/scaling.
Also assess non-tech skills. Can they be effective with marketing, writing, speaking, negotiations?
The point is picking the right strengths for your mission, and the right strengths to complement your own skills is a big deal.