linkedin-post

作者: langchain-ai

根据研究结果或给定主题撰写一篇领英帖子。当被要求创建领英内容、专业帖子或思想……时使用此技能。

npx skills add https://github.com/langchain-ai/langgraph-101 --skill linkedin-post

LinkedIn Post Skill

Format

  • Hook: Start with a bold opening line that grabs attention (this appears before the "see more" cut)
  • Body: 3-5 short paragraphs, each 1-2 sentences
  • Use line breaks between paragraphs for readability
  • Include 1-2 relevant emojis per paragraph (don't overdo it)
  • End with a call-to-action or question to drive engagement
  • Add 3-5 relevant hashtags at the bottom

Tone

  • Professional but conversational
  • Share insights, not just information
  • Use "I" statements and personal perspective where appropriate
  • Avoid jargon unless the audience expects it

Length

  • Ideal: 150-300 words
  • LinkedIn truncates after ~210 characters, so the first line must hook the reader

Template

[Bold hook / surprising stat / question]

[Context -- why this matters]

[Key insight 1]

[Key insight 2]

[Key insight 3 or personal takeaway]

[Call to action / question for engagement]

#hashtag1 #hashtag2 #hashtag3

Example

Most AI agents fail not because of the model -- but because of context management.

After researching the latest agent frameworks, one pattern keeps emerging:
the best agents treat their context window like a scarce resource.

Here's what separates good agents from great ones:

1. They offload intermediate results to a filesystem instead of keeping everything in context
2. They delegate to subagents for isolation -- the main agent only sees summaries
3. They use progressive disclosure -- loading instructions only when relevant

The shift from "bigger context window" to "smarter context management" is where
the real breakthroughs are happening.

What patterns have you seen work best in your agent architectures?

#AIAgents #LangChain #LangGraph #ContextEngineering

来自 langchain-ai 的更多技能

langgraph-docs
langchain-ai
访问LangGraph文档,构建有状态代理和多代理工作流。获取官方LangGraph Python文档,涵盖状态机、基于图的代理设计以及人机协同模式。根据查询类型优先提供相关文档:实现指南解答操作问题,概念页面讲解理论,教程提供端到端示例,API参考提供技术细节。自动选择2–4个最相关的文档URL并检索其内容以回答...
official
langgraph-human-in-the-loop
langchain-ai
暂停图执行以进行人工审查、批准或验证,随后根据其输入恢复执行。需要三个组件:检查点存储器(InMemorySaver 或 PostgresSaver)、配置中的线程 ID 以及 JSON 可序列化的中断负载。interrupt(value) 暂停执行并展示数据;Command(resume=value) 恢复执行并将该值返回给暂停的节点。恢复时,interrupt() 之前的所有代码会重新执行,因此副作用必须具有幂等性(使用 upsert 而非 insert)。支持审批工作流,...
official
web-research
langchain-ai
用于处理与网络研究相关的请求;它提供了一种结构化的方法来进行全面的网络研究
official
langchain-oss-primer
langchain-ai
任何LangChain、Deep Agents或LangGraph代理构建项目都请始终从这里开始。在选择其他技能或编写任何内容之前,这是必需的起点。
official
skill-creator
langchain-ai
创建有效技能的指南,通过专业知识、工作流程或工具集成来扩展代理能力。当用户……时使用此技能。
official
social-media
langchain-ai
根据研究内容起草特定平台的社交媒体帖子,并生成配套图片。支持领英帖子(1300字符,专业语气)和推特/X话题(每条推文280字符,采用1/🧵格式)。需在撰写前将研究任务委托给子代理,随后阅读研究结果以确保准确性和相关性。使用generate_social_image工具自动生成引人注目的社交图片,采用粗体高对比度构图,针对小屏幕进行优化...
official
deep-agents-memory
langchain-ai
为Deep Agents提供可插拔的内存与文件后端,支持临时、持久化和混合路由选项。四种后端类型:StateBackend(线程作用域,临时)、StoreBackend(跨会话持久化)、FilesystemBackend(本地开发时真实磁盘访问)和CompositeBackend(将不同路径路由到不同后端)。FilesystemMiddleware提供六种文件操作工具:ls、read_file、write_file、edit_file、glob、grep。CompositeBackend使用最长前缀匹配进行路由...
official
deep-agents-orchestration
langchain-ai
编排子代理,规划多步骤任务,并对敏感操作要求人工审批。通过任务工具将工作委派给专业子代理;自定义子代理支持独立的工具集和系统提示,而默认的“通用”子代理继承主代理配置。使用write_todos规划并跟踪复杂工作流,将任务组织为待处理、进行中和已完成状态;需要thread_id以实现跨调用的持久化。实现...
official