Reef open-sourced: infrastructure for agents to continuously self-improve from inference-time signals
pliang279 · x · 2026-09-03
- The team behind Reef has open-sourced Reef (github.com/Human-Agent-Society/reef), an infrastructure for continual self-improving agents, built on the idea that "your inference server is secretly a learner."
- Reef combines a model with a harness that evolves from signals generated at inference time, avoiding offline retraining loops.
- The repo is organized into modules (core, harness, runtime, train, observability, scenario, etc.), has 174 stars, and the team welcomes issues and PRs.
More from coding & agent
- Matt Pocock's pre-commit hook caps agent-written files at 5,000 tokens — mattpocockuk · 2026-10-18
- Author Laurence Moroney: write your ugly first draft yourself, then let an LLM edit it — lmoroney · 2026-10-18
- Open-weight TYGER speech decision model makes 80ms turn-taking calls for full-duplex voice agents — solyarisoftware · 2026-10-18
- Devs: Opus 5.5 on medium beats Opus on max — eyishazyer · 2026-10-18
- He Stopped Telling Claude Code to Stop Making Up Timestamps—and Made It Impossible with a PreToolUse Hook — Grimmoner · 2026-10-18
- Cursor Super + Claude Pro + OpenCode: can this stack survive heavy autonomous coding? — DarioAmodeiWasTaken · 2026-10-18