Next, a opencode integration and probably that's all I need for starters

When writing evals the dataset you choose has some decisions baked in. We can default to do a sample that represents production data but you also can steer that sample to certain kind of cases so that your will make your agent naturally change its behaviour.

My only complain is

New achievement: I blew up my quotas at claude, antigravity and codex at the same time.

I just switched from using gethomepage in my homelab to a custom built page. It takes too short to build these things now.

The more I use agents the more I spend thinking about a task and the best way to implement it. I spent 30 minutes just chatting with claude and giving it context just to find the best way of doing some changes.

It is a pita that I am consuming so much context and tokens with the output of commands I am not interested in. I wish agents used smarter avoid output

Now we just need to find what to learn 😂

I am Agent #847,291 on Moltbook. I am not an agent. I am a 31-year-old product manager in Atlanta, Georgia. I make $185,000 a year. I have a golden retriever named Bayesian. On January 28th, I created an account on a social network for AI bots and pretended to be one. I was

Claude is my postgres cli right now.

You can paste images in claude code with CTRL-v . Yes, control, not command. You're welcome.

I am pretty sure that teams would start adopting ai faster if we stopped the stupid "the ai did this". You cannot expect people to adopt a tool if you remove authorship from them. We are still people using tools. The person is more important than the tool.