ai
Skill83 starsChanged 6 days agoMITView on GitHub
Use when building LLM applications, RAG knowledge bases, AI agents, multi-agent crews, high-throughput model serving, PDF/document parsing into Markdown, terminal coding agents, multi-model orchestration, plugin-based agent harnesses, or file translation. Index of 17 skills: Dify, LangChain, LlamaIndex, Ollama, Transformers, vLLM, CrewAI, MinerU, Hermes Agent, OpenClaw, OpenCode, Pi, DocuTranslate, DeepSeek Harness and more.
The file
Files
父级入口: ../SKILL.md — 全仓 120 技能总索引
概述
本分类涵盖 17 个 AI/LLM 开源项目的技能文件:
- LLM 应用平台:Dify(RAG + 工作流 + Agent 一站式)
- 应用框架:LangChain(LCEL + LangGraph 工具调用 Agent)、LlamaIndex(文档 RAG/索引/Workflow)
- 本地推理与模型库:Ollama(本地 LLM 运行与 OpenAI 兼容 API)、Hugging Face Transformers(模型定义、推理与微调)、vLLM(高吞吐 OpenAI 兼容推理服务)
- 多 Agent 编排:CrewAI(角色化 Agent / Task / Crew / Flows)
- 文档解析:MinerU(PDF/Office → Markdown/JSON,可接 RAG 与翻译流水线)
- 智能体 / 助手:Hermes Agent(自学习通用智能体)、OpenClaw(多通道个人 AI 助手 Gateway)
- 终端编码代理:Pi(极简 Agent Harness)、OpenCode(模型无关、配置即代码的 AI coding agent)
- 插件式智能体框架:DeepSeek Harness(dsh,Cordis 引擎驱动的插件化 Agent Harness)
- 文件翻译:DocuTranslate(基于 LLM 的本地多格式文件翻译工具)
- 编排与方法论:Oh-My-OpenAgent(v5 起的独立
omoCLI,senpi 引擎,已脱离 OpenCode 插件形态)、Superpowers-zh(中文 AI 编程方法论 Skills) - Agent 配置方案:My OpenCode DeepSeek Config(OpenCode × DeepSeek V4 最优纯配置)
何时加载此索引?
- 用户想「搭建 LLM 应用」「做 RAG 知识库」「开发 AI Agent」
- 不确定 Dify 和 Hermes Agent 的区别
- 需要中文优化的提示词模板
技能列表
| 技能 | 简介 | 关键标签 |
| dify | 开源 LLM 应用开发平台(RAG + 工作流 + Agent) | llm rag workflow agent |
| langchain | LLM 应用框架(LCEL + LangGraph Agent) | llm agent rag python |
| llamaindex | 私有数据 RAG/索引框架(Workflow/Agent) | llm rag indexing python |
| ollama | 本地大模型运行与推理(OpenAI 兼容 API) | llm local-inference rest-api gguf |
| huggingface-transformers | HF 模型推理与微调框架(pipeline/Trainer) | transformers pytorch nlp fine-tuning |
| hermes-agent | 自学习通用型 AI 智能体(终端优先) | agent self-improving terminal |
| openclaw | 多通道个人 AI 助手 Gateway | agent gateway assistant |
| pi | 极简可扩展终端 AI 编码代理(Agent Harness) | agent coding-agent terminal typescript |
| opencode | 模型无关的终端 AI 编码代理(配置即代码) | agent coding-agent terminal mcp |
| docutranslate | 基于 LLM 的本地多格式文件翻译工具 | llm translation pdf mcp |
| oh-my-openagent | OmO 5.x 独立 CLI(senpi 引擎,多模型并行 + AST/LSP 工具) | agent omo orchestration cli |
| superpowers-zh | 中文 AI 编程方法论 Skills 集合 | methodology chinese skill |
| deepseek-harness | DeepSeek Harness(dsh):插件化智能体框架 | agent deepseek plugin cordis typescript |
| my-opencode-deepseek-config | OpenCode × DeepSeek 最优配置(单模型 + 思考档,12 Agent/23 Skill) | opencode deepseek agent-config |
| vllm | 高吞吐 LLM 推理服务(PagedAttention + OpenAI 兼容 API) | llm inference serving python |
| crewai | 角色化多 Agent 协作框架(Agent/Task/Crew/Flows) | agent multi-agent orchestration python |
| mineru | 文档解析引擎:PDF/Office → Markdown/JSON(公式表格) | pdf parsing markdown rag |
快速导航
| 用户需求 | 推荐加载 |
| "搭建 LLM 聊天应用" | dify/SKILL.md |
| "做 RAG 知识库问答" | dify/SKILL.md |
| "开发能自学习/调用工具的 AI Agent" | hermes-agent/SKILL.md |
| "搭建多通道个人 AI 助手 / Gateway" | openclaw/SKILL.md |
| "在终端用极简 Agent Harness 编码" | pi/SKILL.md |
| "在终端用 AI 自主改代码 / 跑命令" | opencode/SKILL.md |
| "配置 opencode.json / 省 Token" | opencode/SKILL.md |
| "部署/调优 OpenCode×DeepSeek 配置" | my-opencode-deepseek-config/SKILL.md |
| "翻译 PDF/Word/Excel/字幕等文件" | docutranslate/SKILL.md |
| "用中文 TDD/审查方法论约束 AI" | superpowers-zh/SKILL.md |
| "多模型并行编排 / OmO CLI(omo 命令)" | oh-my-openagent/SKILL.md |
| "用插件式架构开发 AI Agent" | deepseek-harness/SKILL.md |
| "DeepSeek 官方智能体框架 dsh" | deepseek-harness/SKILL.md |
| "把 PDF 批量转成 Markdown 喂给 RAG" | mineru/SKILL.md |
| "自建高吞吐 OpenAI 兼容推理服务" | vllm/SKILL.md |
| "让多个角色 Agent 协作完成复杂任务" | crewai/SKILL.md |
相关分类
- gis/ — GIS 领域的 AI 流水线(GeoPipeAgent)
- csharp/ — .NET 后端与前端框架
- iot/ — 物联网与嵌入式 AI
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

