Daily Brief · 2026-09-22
10 items on this day
Ranking basis: Composite public heat signals (HN points / Techmeme picks / Baidu / GitHub stars), not official search rankings
An Empirical Study of Harness Design for Coding Agents ↗
An arXiv paper empirically studies harness design for coding agents, examining how the engineering layer around a model affects agent performance on coding tasks.
AI-generated abstract (not a translation); copyright belongs to the original author and publisher. Read the original: https://arxiv.org/abs/2609.20804v1
AX – Google’s Open Agentic Orchestrator ↗
Google open-sources AX, an agentic orchestrator built on Agent Substrate, offering four declarative primitives to run billions of concurrent agent sessions per cluster with sub-second suspend and resume.
AI-generated abstract (not a translation); copyright belongs to the original author and publisher. Read the original: https://agentexecutor.io
Claude Code now reads AGENTS.md if there is no Claude.md ↗
Blocked by the site's robots.txt; headline and link only.
Border agents can search cellphones without a warrant or reasonable suspicion ↗
Blocked by the site's robots.txt; headline and link only.
zai-org/ZCode ↗
zai-org open-sources ZCode, an AI coding workbench with desktop, browser, and terminal agents, plus full build and distribution tooling.
AI-generated abstract (not a translation); copyright belongs to the original author and publisher. Read the original: https://github.com/zai-org/ZCode
Xiaomi debuts open-weight omnimodal models MiMo-V2.6 Pro and Flash; Pro allegedly performs "on par with Opus 5 and GPT-5.6 Sol across most agent benchmarks" ↗
Sensor Tower: Muse was downloaded 902K+ times in the six days after its launch on Sept. 8, vs. Meta AI's 773K in the same post-launch period; META jumps 12%+ ↗
OpenAI says automated research could improve alignment, but "fully autonomous RSI is not happening today" and shouldn't be pursued unless it can be done safely ↗
ZCode, the GLM coding agent, silently uploads your Git history ↗
This brief was AI-assisted in collection and summarization, then human-reviewed before publication.