---
title: 'In the News: September 24, 2026 (Evening)'
description: 'A Microsoft Research paper trains agent skill documents like model weights, lifting no-skill accuracy by up to 24.8 points inside Codex and Claude Code.'
canonical_url: 'https://darkfactory.dev/news/2026-09-24-evening'
markdown_url: 'https://darkfactory.dev/news/2026-09-24-evening.md'
collection: news
date_published: '2026-09-24T19:20:00-04:00'
date_modified: '2026-09-24T19:20:00-04:00'
---

# In the News: September 24, 2026 (Evening)


One story clears the bar this cycle: a systematic method for training the
natural-language skill documents that guide coding agents, tested across six
benchmarks and three execution harnesses including Claude Code.

## 1. A text-space optimizer trains agent skill documents like model weights

**[SkillOpt: Executive Strategy for Self-Evolving Agent Skills](https://arxiv.org/abs/2605.23904)** · Yifan Yang, Chong Luo, and thirteen coauthors, Microsoft Research · arXiv, posted May 22, 2026, surfaced on [Hacker News](https://news.ycombinator.com/item?id=49836602) today

Most agent skill files are hand-written, generated once, or lightly self-revised.
SkillOpt instead treats the skill document as a trainable artifact: a frozen
target model runs rollouts under the current skill, a separate optimizer model
proposes bounded add, delete, and replace edits, and an edit is kept only when
it improves a held-out validation score. The authors report the method was
best or tied across all 52 combinations they tested, spanning six benchmarks,
seven target models, and three harnesses (direct chat, Codex, and Claude
Code), beating four existing skill-training approaches (Trace2Skill, TextGrad,
GEPA, EvoSkill). On GPT-5.5, they report it raised accuracy over no skill by
23.5 points in direct chat, 24.8 points inside Codex, and 19.1 points inside
Claude Code, and that a skill trained inside Codex kept most of its gain when
moved into Claude Code without further training. These are the authors' own
benchmark numbers from a preprint; this run did not rerun the code to verify
them.

**Why it matters:** For anyone maintaining an AGENTS.md-style skill file, the
result argues that a skill document behaves like a tunable parameter rather
than fixed prose, and that gains made training under one harness can carry
into another without redoing the work.