Post B76Na7M08SJbSbzKng by TyggBiddies@clew.live
 (DIR) More posts by TyggBiddies@clew.live
 (DIR) Post #B76N7XkAwIlEuRfcSu by verita84_64@poster.place
       2026-06-07T21:05:02.011830Z
       
       1 likes, 0 repeats
       
       Linux Chads, automate Tedious CLI work with bash aliases and opencode!  In this demo, my goal is to copy stickers to my Matrix server for later use. The stickers command calls opencode to do the bullshit janny work. opencode is connected to PosterChanAI, self-hosted agentic powerhouse #ai #linux
       
 (DIR) Post #B76Na7M08SJbSbzKng by TyggBiddies@clew.live
       2026-06-07T21:10:14.949807Z
       
       1 likes, 0 repeats
       
       @verita84_64 have you tested silver trajectories on FreeBSD? Jails are super annoying to work with but they can do wonders; same with OpenBSD subnetting.
       
 (DIR) Post #B76NgAMJsKO8WC1NrM by verita84_64@poster.place
       2026-06-07T21:11:20.380256Z
       
       0 likes, 0 repeats
       
       never heard of it so no @TyggBiddies
       
 (DIR) Post #B76OPWGXH0bRUYwTFQ by TyggBiddies@clew.live
       2026-06-07T21:19:32.393129Z
       
       1 likes, 1 repeats
       
       @verita84_64 Oh, silver trajectories are part of AI agent development lifecycle. It basically means that you let an agent run the same set of prompts on several models and choose the most accurate one among them by comparing your rubric with their result. There are measurable ways to evaluate disagreements like Cohen's kappa or Krippendorf's alpha for fine tuning but usually a decent set of rules will do.They're called 'silver' because you can expect failures and still use them unlike a golden trajectory experiment where you have to hand pick the Ground of Truth by hand and evaluate your agent against it.
       
 (DIR) Post #B76b0Q5EQ9V6Ty15SC by verita84_64@poster.place
       2026-06-07T23:40:39.393914Z
       
       0 likes, 0 repeats
       
       @grey