Claude Code 한도 50% 상향·Grok·Pi 통합과 MACE 논문 등 에이전트·플랫폼 변화

ClaudeDevs

@ClaudeDevs

We're also keeping Claude Code weekly limits 50% higher, now through August 19, for all Pro, Max, Team, and seat-based Enterprise users.

Elon Musk

@elonmusk

Try Grok Build https:// x.ai/cli

Grok

@grok

Use any paid SuperGrok or X Premium subscription inside the Pi agent

AI agent coordination is going to be the next leap in AI. This paper talks about the problem of multi-agent exploration and how to improve it. (bookmark it) Let's break it down: The default assumption in multi-agent work is that capable LLMs, placed together, will probe each other and self-organize into good coordination. New research shows they do not. Modern LLM agents fail to explore each other. They fall into myopic, polarized interaction patterns that raise regret and leave coordination well short of optimal. The authors formalize this as a Multi-Agent Exploration problem, a partially observable stochastic game where agents must probe peers to infer capability, then introduce MACE, a lightweight framework that promotes exploration through structured peer selection. Paper: https:// arxiv.org/abs/2607.11250 Learn to build effective AI agents in our academy: https:// academy.dair.ai

You can use K3 as a teacher model (with logits!) to train up smaller dense models. It’s actually feasible even to use it to create a trillion token full pre and post training dataset for even better models.

Was thinking of trying different agents & harnesses like @cognition & @opencode with Kimi K3 What are people’s favourites these days

Santiago

@svpino

Loop Engineering is dead. Long live Graph Engineering!

Jerry Liu

@jerryjliu0

1. build a workflow graph on top of the agent harness 2. dynamically create that graph through an outer agent loop ....repeat life is a loop (and a graph)