#workflows

564 stories · 28 in 7 days · 106 in 30 days
Stories, products and papers tagged "workflows", newest first.
Nov 13
How-toBuildersOfficial08:25
Chinese contentGitHub 实战指南:构建一次即可在任何智能体上运行的 Agent Skills

GitHub 官方这篇教程讲清了 Agent Skills 的写法和分层思路:工作流写成文件复用,工具和权限留在宿主侧。如果你在 Copilot 或其他智能体上重复做同类任务,照着做一次就能到处跑。

Primary sourceGitHub Blog@Andrea GriffithsSource
Nov 12
How-toBuildersMulti-source16:02
Chinese content用好 Coding Agent 的关键:把协作当管理来做,附 12 条实践

这篇把和 Claude Code、Cursor 这类编程 Agent 的日常协作拆成了管理流程:选任务、写要求、跟进度、验收上线。里面 12 条实践都偏可操作,不是空谈。每天让 Agent 写代码的人照着做,能少走不少弯路。

Event
AI Will@FinanceYF5Source
How-toBuildersMulti-source15:33
Chinese content用 Fable 5.5 以纯代码生成83秒动画短片,8组Agent并行3小时内完成

有人拿 Fable 5.5 不用视频模型,纯靠8组Agent写 Three.js 代码渲染出一部83秒短片,39个镜头1992帧,最后就一个HTML文件。比 Sora 按帧生成省钱得多,想精确控制画面的开发者可以照这个流程试试。

Event
AI Will@FinanceYF5Source
Nov 11
Chinese contentReTrace:将工具调用 Agent 蒸馏为无需 LLM 规划的代码工作流

做流程自动化的团队可以看看这篇。思路挺聪明:不每次跑都让 LLM 决策,而是从 Agent 日志里挖出分支条件,编译成纯代码工作流。速度从 170 秒掉到 1.3 秒,成本降 24%,准确率只掉两三个点。跑固定 SOP 的场景很划算。

arXiv cs.AI@Kaiyuan Hou et al. (9)Source
Nov 10
How-toBuildersSelected19:38
LangChain explains eval-driven development for improving agents

The counterintuitive point is that aggregate scores can rise while previously correct behaviours regress, so per-behaviour comparison against a baseline is the only safe way to ship agent changes. Builders maintaining agents in production, especially those wiring in new tools, can apply this workflow to catch regressions early. The example also shows that many 'model failures' are just badly scoped tool descriptions — a cheap fix teams often overlook.

shao__meng@shao__mengSource
Chinese contentWarp 的 Zach Lloyd 演示:智能体连续失败 37 次后主动提交 PR 修复自身工作流

Warp 的 Zach Lloyd 放出的演示挺有意思:智能体同一错误摔了 37 次,干脆自己提了个 PR 修工作流。做 Agent 的朋友可以看看这种"失败次数触发自我修复"的设计思路,他这周还会在旧金山 AIE CODE 现场讲。

AI Engineer@aiDotEngineerSource
How-toBuildersMulti-source05:22
Chinese contentAnthropic 的 Alex Albert 分享新模型发布后的提示词清理方法

Anthropic 的 Alex Albert 透露他每次换新模型就用这条提示词全项目体检,让模型自己找出过时的补丁指令。写Claude或Cursor提示词的人可以抄走直接用,几分钟扫一遍就知道哪些规则该删。

Event
宝玉@doteySource
Nov 9
Nov 8
How-toBuildersSelected12:32
Brian Armstrong's multi-agent setup: lead agents and one queue

The pattern matters more than the tools: one queue plus persistent memory beats juggling dozens of chat sessions, because coordination overhead, not model quality, becomes the bottleneck with frontier models. Engineering leads and solo builders can replicate the structure with any agent framework by defining lead agents, sub-agents and a single decision inbox. All changes still pass human review and security checks, a useful guardrail template for teams adopting similar setups.

elvis@omarsar0Source
Nov 7
Nov 6
Nov 5