Video wird geladen...
Video konnte nicht geladen werden
Creator of Claude Code: “I’m not just prompting agents anymore. I’m building a harness with graphs and loops - and the agents inside it are building my agent. In 3–6 months, we’ll all be engineering harnesses for our agentic systems.” In this 10-minute talk, Boris Cherny reveals what the... show more
72,217 Aufrufe • vor 3 Tagen •via X (Twitter)
6 Kommentare

catmanvor 3 Tagen
A harness is like a workshop jig: graphs and loops constrain the agent’s work so it can build and improve components without losing the process.

beamnxw ./vor 3 Tagen
the model is becoming the worker and the harness becomes the company

Robert Corbin | Shadowfetchvor 3 Tagen
The practical win is making each run verify its own work before moving on, so a harness catches skipped tests and wrong tool calls instead of hiding them.

DHRUV BANSALvor 1 Tag
once you're running a few thousand agents overnight the harness is mostly deciding what you never read. with nobody awake to check it

kozh ./vor 3 Tagen
The main thing is to set the task

安叫兽|Bird🕊️ 🔶 BNBvor 3 Tagen
以后调 bug 得先问一句,是哪个代理写的代理?
