UC Berkeley's Actor-Gatherer: open-source agent context management, +32% accuracy
adityagp · x · 2026-10-21
UC Berkeley researchers open-sourced Actor-Gatherer, a bolt-on framework for AI agent context management. It separates task-solving from context management: an Actor agent executes tasks while a Gatherer component proactively decides what context to inject, observing the Actor, other agents, and environment changes.
- Targets long-running sessions, cross-agent context sharing, and changing environments where history exceeds LLM context windows
- +32% accuracy over state-of-the-art for a long-running single agent; +31% with parallel agents in changing environments
- Integrates with Claude Code, Codex, Pi, and OpenCode
Code, docs and blog are available.
Related event: UC Berkeley Open-Sources Actor-Gatherer for Multi-Agent Context Management(2 posts)→
More from coding & agent
- Salesforce AI's SRD aligns agent foresight with hindsight at zero test-time cost — lupantech · 2026-10-21
- Claude autonomously shrinks attack surface of 25 OSS codebases, hardens OpenSSH — chrisrohlf · 2026-10-21
- WildAWM: fully open-source agent task synthesis yields +10.58 OOD gains for Qwen3.6-35B — XuCanwen · 2026-10-21
- Developing Taste: Emil Kowalski on why judgment beats tools in the AI era — rseroter · 2026-10-21
- MCP in action: Claude Opus hands off asset library videos to Runneth, launching Friday — jacob_posel · 2026-10-21
- Mastercard taps Coinbase's x402-based infrastructure to let AI agents pay for digital services — MurrLincoln · 2026-10-21