bookmarks

@PrimeIntellect

Prime Intellect · 83,169 followers

Open Superintelligence Stack

2 bookmarks

View all →

Introducing Prime Agent: A self-improving RLM harness for coding and long-running autonomous tasks. Designed to be both token-efficient and expressive through programmatic tool calling, context as a variable, multi-agent messaging, and a self-modifiable harness state.↗