- 发布日期
本文深入探讨如何在 LangChain Deep Agents 架构中集成 Skills 技能系统。文章对比了 CLI 与 Library 的实现差异，详细介绍了通过 CompiledSubAgent 封装技能、利用 Progressive Disclosure 动态加载以及 PythonREPLTool 执行环境的完整技术方案。
Claude-skills
全部文章
- ai-agent (45)
- python (20)
- mcp (14)
- langgraph (13)
- rag (11)
- llm (10)
- 读书笔记 (10)
- langchain (10)
- 软件架构 (8)
- ai (7)
- 大语言模型 (7)
- workflow (7)
- ai-coding (6)
- 多智能体系统 (5)
- engineering (5)
- security (5)
- ontology (5)
- 状态机 (4)
- context-engineering (4)
- 知识图谱 (4)
- devops (4)
- agent-evaluation (4)
- 数据分析 (3)
- dify (3)
- deepseek (3)
- 深度研究 (3)
- claude (3)
- skills (3)
- agentic-ai (3)
- 架构演进 (3)
- agent-skills (3)
- owl (3)
- 生成式ai (2)
- 强化学习 (2)
- 人工智能 (2)
- 机器学习 (2)
- javascript (2)
- ddd (2)
- hyde (2)
- llm应用 (2)
- java (2)
- 依赖管理 (2)
- 虚拟环境 (2)
- 包管理 (2)
- python工具链 (2)
- 提示工程 (2)
- deerflow (2)
- 企业级ai (2)
- 领域建模 (2)
- a2a (2)
- 代理框架 (2)
- 教程 (2)
- deep-agents (2)
- multi-agent (2)
- ai-architecture (2)
- anthropic (2)
- react (2)
- content-extraction (2)
- knowledge-base (2)
- graphrag (2)
- owasp (2)
- knowledge-graph (2)
- enterprise-architecture (2)
- agent-architecture (2)
- knowledge-management (2)
- semantic-layer (2)
- data-governance (2)
- 描述逻辑 (2)
- 知识表示 (2)
- harness (2)
- linkml (2)
- 本体建模 (2)
- generative-ai (1)
- 乱翻书 (1)
- 蒙特卡罗方法 (1)
- 数值计算 (1)
- 随机模拟 (1)
- q-learning (1)
- 马尔可夫决策过程 (1)
- 环境搭建 (1)
- github-pages (1)
- blog (1)
- eslint (1)
- 前端工具 (1)
- 代码质量 (1)
- 静态分析 (1)
- cola (1)
- 分层架构 (1)
- 领域驱动设计 (1)
- 微服务 (1)
- java架构 (1)
- prompt (1)
- 随手记 (1)
- google (1)
- prophet (1)
- facebook (1)
- 时间序列预测 (1)
- 价格预测 (1)
- 加法模型 (1)
- 趋势分析 (1)
- 季节性分析 (1)
- clickhouse (1)
- echarts (1)
- 数据可视化 (1)
- ai应用开发 (1)
- llm工具集成 (1)
- rag技术 (1)
- rse (1)
- 自我rag (1)
- 自适应rag (1)
- 重排序器 (1)
- 图rag (1)
- 层次rag (1)
- crag (1)
- 检索增强生成 (1)
- ai开发 (1)
- spring-boot (1)
- ai集成 (1)
- 开发工具 (1)
- spring-ai (1)
- poetry (1)
- pyprojecttoml (1)
- uv (1)
- streamable-http (1)
- 异步编程 (1)
- 服务器开发 (1)
- 客户端开发 (1)
- asgi (1)
- 分布式系统 (1)
- 实时通信 (1)
- mem0 (1)
- openmemory (1)
- ai记忆 (1)
- 向量存储 (1)
- 本地部署 (1)
- 记忆系统 (1)
- 12-factor-agents (1)
- 生产级agent (1)
- ai工程 (1)
- 最佳实践 (1)
- baml (1)
- ai工作流 (1)
- 代理 (1)
- 类型安全 (1)
- boundaryml (1)
- ai架构 (1)
- fastapi (1)
- 工作流编排 (1)
- 智能体协作 (1)
- embabel (1)
- jvm (1)
- spring (1)
- kotlin (1)
- 智能规划 (1)
- rod-johnson (1)
- ooda循环 (1)
- open-deep-research (1)
- 多代理系统 (1)
- 架构设计 (1)
- deep-research (1)
- 研究报告 (1)
- nextjs (1)
- vercel (1)
- 多模型架构 (1)
- typescript (1)
- fastmcp (1)
- 工具集成 (1)
- 模型上下文协议 (1)
- subgraph (1)
- 架构 (1)
- 管理 (1)
- 开发 (1)
- 定律 (1)
- 软件工程 (1)
- 产品管理 (1)
- 创业方法论 (1)
- design-sprint (1)
- 产品战略 (1)
- deepagents (1)
- 工具包 (1)
- 代理运行时 (1)
- tavily (1)
- 任务规划 (1)
- 上下文管理 (1)
- chrome-devtools (1)
- 浏览器自动化 (1)
- ai编程助手 (1)
- trae-ai (1)
- puppeteer (1)
- 性能监控 (1)
- 调试工具 (1)
- web开发 (1)
- agents-20 (1)
- 人类在环 (1)
- 持久化 (1)
- 长期记忆 (1)
- 子代理 (1)
- ai工程化 (1)
- kubernetes (1)
- kind (1)
- helm (1)
- macos (1)
- supervisor (1)
- agentic-patterns (1)
claude-skills (1)
- pdf (1)
- microsandbox (1)
- daytona (1)
- sandbox (1)
- swarm (1)
- multi-hyde (1)
- adaptive-hyde (1)
- sql (1)
- 架构框架 (1)
- finops (1)
- 混合智能 (1)
- 深度可观测性 (1)
- 反射型智能体 (1)
- 多智能体编排 (1)
- tool-selection (1)
- multi-agent-systems (1)
- orchestration (1)
- evaluation (1)
- testing (1)
- metrics (1)
- llm-as-judge (1)
- best-practices (1)
- monitoring (1)
- observability (1)
- opentelemetry (1)
- sre (1)
- improvement-loops (1)
- feedback-pipelines (1)
- experimentation (1)
- continuous-learning (1)
- maestro (1)
- prompt-injection (1)
- defense-in-depth (1)
- human-agent-collaboration (1)
- guided-autonomy (1)
- governance (1)
- trust (1)
- superpowers (1)
- tdd (1)
- spec-driven-development (1)
- cognee (1)
- memory (1)
- arckit (1)
- architecture-governance (1)
- architecture (1)
- rdf (1)
- shacl (1)
- enterprise-ai (1)
- dikwp (1)
- tool-call (1)
- function-calling (1)
- authorization (1)
- okf (1)
- code-review (1)
- amazon-bedrock (1)
- graphify (1)
- agent-harness (1)
- foundry-iq (1)
- 知识检索 (1)
- 多智能体 (1)
- 工程实践 (1)
- postgresql (1)
- webvowl (1)