I have a couple of projects I'm working on right now don't want to say what they are, because publicly searchable. It's a small tool in a boring area. But right now I'm not using 6.1 to build anything, I'm using it to pull a bunch of stuff from my cloudflare api, because there was an issue yesterday where a bot was making a request that should have been covered by my regex syntax were getting edge 301s when they should have been mitigated. Actually its saying it figured it out in its thought chain, it is trying to say that for https requests they hit botfight mode and for https, but I tried it yesterday from my terminal using the exact same protocol they were using and got 403, so idk if I believe it yet, I'll have to read the report. Then I'm also having it pull what wasn't mitigated for the day because there is a graph ql error right now on my dashboard. Also some fag is using cloudflare workers as probe bots and putting my application address as its user agent and its pissing me the fuck off, even though my rules cover the probes and they don't actually apply to me. It's spamming my logs though.
Agents are too expensive to use directly in trading imo. its burning so many tokens just researching a market. I'd set something up using a decision engine like Jev or an open source alternative that you can run for free (although Jev is the most accurate one). Ask it to predict when Kalshi is wrong.