Tag
This paper proposes decomposing the agent harness into eight functional components and a portable, versioned specification framework that records which model ran in which harness, addressing score variability and attack-surface documentation gaps not covered by existing model/agent cards.
A quote from Kyle Kingsbury (Jepsen) arguing that LLMs erode trust in software, documentation, and the organizations producing them, and asking how maintainers can demonstrate software quality in the age of LLMs.
An explainer on choosing among groff/troff macro packages (ms, mm, me, mom), covering what troff/groff are, the benefits of plain-text typesetting over word processors, and when to use each macro package.
Recommended an open-source project titled “Network Proxies and VPN Applications Explained,” covering the fundamentals of proxies, tunneling, and VPNs, as well as network blocking mechanisms such as IP blocking, DNS poisoning, TCP RST, and SSL blocking, with diagrams and real-world examples.
The authors measure how well project rules reach coding agents, finding Claude Code captures only 32% of needed rules without docs (54% with, 70% with their Segmently Modules setup), while code-indexing tools like Serena, Trail and Graphify add little. They release 'the stand', an open-source MIT-licensed context-recall benchmark kit that runs as a Claude Code plugin on real past tasks.
Spacesheep (YC F26) launches on Product Hunt, turning any AI agent's output into shareable interactive docs, live dashboards, mockups, or apps for team collaboration.
Falconer launched to give production AI agents a unified, auto-updated knowledge base combining internal docs and public documentation, backed by GP, BoxGroup, SV Angel and angels from Anthropic, Stripe and Cursor.
SVPino argues companies already have enough model intelligence and instead need better knowledge organization, highlighting Falconer, a tool that links code with a company knowledge base and auto-updates outdated docs on pull request merges.
Nessa Labs 分享了 nessa-docs 开源工具：让编码代理在项目 docs/state 中生成 Markdown 系统地图与流程图，并通过本地 Web 应用以可拖拽、可缩放的交互图表浏览，帮助开发者理解和维护代码库状态。
Phil Schmid 介绍 Google 新推出的 Gemma 官方文档站点，涵盖模型能力、微调指南、推理优化以及模型工作原理的教育内容，全部集中在一个平台上。
Dev Skills 是一个为编码代理设计的可复用工作流技能库，把不同代理的测试、提交、失败排查和记忆等决策规范化为 Markdown 指令，灵感来自万智牌（Magic: The Gathering）的组织方式。
Matt Pocock asks how developers are finding the GLOSSARY.md approach from /grill-with-docs, noting that investing more in the glossary makes him more productive and helps the AI agent 'speak his language'.
The README team introduces Restless, a tool that makes APIs agent-friendly by recording AI agent requests via `npx restless init` and having error responses guide agents toward fixes. With ~72% of docs traffic already coming from agents, the launch argues companies should treat AI agents as their primary users.
作者反思如何用"实现决策记录"（IDRs）提升 LLM 代理的输出质量，指出有效的 IDR 必须围绕意图与决策展开，否则只是用英文复述代码差异。
探讨多语言(polyglot)文档站点的构建难题与两大整合策略——转换(transform)与嵌入(embed),并以 Breathe 和 Doxygen/Sphinx 在 pigweed.dev 的实践为案例,分析各自的权衡与痛点。
The author paid volunteers €25 each to walk through ActivityBot's README aloud on video calls, uncovering confusing instructions and iteratively improving the docs — arguing real human feedback beats LLM-simulated users for testing documentation.
ste-zh 是一个开源 Agent Skill，借鉴 ASD-STE100 简化技术英语规范，约束 AI Agent 用清晰、无歧义的中文汇报工作结果，支持通过 skills CLI 或手动安装到 Claude Code 等环境。
一条关于使用 Codex 时增加「费曼报告」流程的提示词技巧：要求 Codex 在完成任务每个阶段后，用通俗语言向完全不了解项目的人汇报，并用 PlantUML 绘制进度图。
Fenng mentions he had already borrowed and applied the concept of STE (ASD-STE100 Simplified Technical English) to his Chinese technical documentation writing Skill, which has since iterated to a new version adding rules inspired by STE to reduce ambiguity and operational errors, and has now reached nearly 500 stars.
Author hylarucoder released the hai-simplified-technical skill, adapting Simplified Technical English (STE) from the aviation maintenance documentation standard ASD-STE100 into 46 Chinese writing rules, along with a check.py validation script to help AI coding agents produce unambiguous, actionable technical documentation.