整理文件路径:Technical→AI/
This commit is contained in:
25
CLAUDE.md
25
CLAUDE.md
@@ -9,6 +9,7 @@
|
||||
|
||||
- 所有输出必须使用**简体中文**
|
||||
- 专有名词允许保留英文,但首次出现必须附带中文解释
|
||||
- 如果原始文件名是中文,则source页面的名称尽量用中文,不要用拼音表示, 如果有特殊字符可以忽略
|
||||
- 禁止中英混合句(术语除外)
|
||||
- 不允许输出纯英文总结或分析
|
||||
|
||||
@@ -46,12 +47,12 @@ Transformer(变压器模型,一种基于注意力机制的神经网络架构
|
||||
|
||||
# Slash Commands(Claude Code)
|
||||
|
||||
| Command | 使用方式 |
|
||||
|---|---|
|
||||
| `/wiki-ingest` | `ingest raw/your-file.md` |
|
||||
| `/wiki-query` | `query: 你的问题` |
|
||||
| `/wiki-lint` | `lint the wiki` |
|
||||
| `/wiki-graph` | `build the knowledge graph` |
|
||||
| Command | 使用方式 |
|
||||
| -------------- | --------------------------- |
|
||||
| `/wiki-ingest` | `ingest raw/your-file.md` |
|
||||
| `/wiki-query` | `query: 你的问题` |
|
||||
| `/wiki-lint` | `lint the wiki` |
|
||||
| `/wiki-graph` | `build the knowledge graph` |
|
||||
|
||||
---
|
||||
|
||||
@@ -107,13 +108,11 @@ last_updated: YYYY-MM-DD
|
||||
---
|
||||
|
||||
# Ingest Workflow(摄取流程)
|
||||
**重要** 请严格按照摄取流程进行操作,每分析一个页面必须要创建/更新source page,entity, concept等。不可遗漏!
|
||||
|
||||
触发方式:
|
||||
- `/wiki-ingest`
|
||||
- 或:`ingest <file>`
|
||||
|
||||
---
|
||||
|
||||
## 执行步骤(严格顺序)
|
||||
1. 使用 Read 工具完整读取 source 文档
|
||||
2. 读取 `wiki/index.md` 和 `wiki/overview.md`
|
||||
@@ -135,11 +134,10 @@ title: "Source Title"
|
||||
type: source
|
||||
tags: []
|
||||
date: YYYY-MM-DD
|
||||
source_file: raw/...
|
||||
---
|
||||
|
||||
## Source Files
|
||||
- [[SourceFiles]]
|
||||
## Source File
|
||||
- [[raw/...]]
|
||||
|
||||
## Summary
|
||||
- 核心主题:
|
||||
@@ -278,7 +276,8 @@ date: YYYY-MM-DD
|
||||
- 断链
|
||||
- 冲突
|
||||
- 过期内容
|
||||
- 缺失实体
|
||||
- 缺失Entity
|
||||
- 缺失Concept
|
||||
- 知识空白
|
||||
|
||||
---
|
||||
|
||||
BIN
raw/Technical/.DS_Store
vendored
BIN
raw/Technical/.DS_Store
vendored
Binary file not shown.
@@ -1,70 +0,0 @@
|
||||
---
|
||||
title: Templater Obsidian Plugin
|
||||
source:
|
||||
author: shenwei
|
||||
published:
|
||||
created:
|
||||
description:
|
||||
tags: [obsidian, plugin]
|
||||
---
|
||||
|
||||
|
||||
#obsidian #plugin
|
||||
|
||||
# Templater Obsidian Plugin
|
||||
|
||||

|
||||
|
||||
[Templater](https://github.com/SilentVoid13/Templater) is a template plugin for [Obsidian.md](https://obsidian.md/). It defines a templating language that lets you insert variables and functions results into your notes. It will also let you execute JavaScript code manipulating those variables and functions.
|
||||
|
||||
## Documentation
|
||||
|
||||
Check out the complete [documentation](https://silentvoid13.github.io/Templater/) to start using [Templater](https://github.com/SilentVoid13/Templater).
|
||||
|
||||
## Warning
|
||||
|
||||
[Templater](https://github.com/SilentVoid13/Templater) allows you to execute arbitrary JavaScript code and system commands.
|
||||
|
||||
It can be dangerous to execute arbitrary JavaScript code or system commands from untrusted sources. Only run code / commands that you understand, from trusted sources.
|
||||
|
||||
## Template Showcase / Questions / Ideas / Help
|
||||
|
||||
Go to the [discussion](https://github.com/SilentVoid13/Templater/discussions) tab to ask and find this kind of things.
|
||||
|
||||
Don't be shy and share your templates created using [Templater](https://github.com/SilentVoid13/Templater) in the [Template Showcase](https://github.com/SilentVoid13/Templater/discussions/categories/templates-showcase) category. Use [gists](https://gist.github.com/) to share the template file.
|
||||
|
||||
## Resources
|
||||
|
||||
A list of useful resources about [Templater](https://github.com/SilentVoid13/Templater):
|
||||
|
||||
- @GitMurf quick demo `How to setup and run your first Templater JS script`: [https://github.com/SilentVoid13/Templater/discussions/187](https://github.com/SilentVoid13/Templater/discussions/187)
|
||||
- @shabegom `How To Use Templater JS Scripts`: [https://shbgm.ca/blog/obsidian/how-to-use-templater-js-scripts](https://shbgm.ca/blog/obsidian/how-to-use-templater-js-scripts)
|
||||
- @chhoumann Templates showcase: [https://github.com/chhoumann/Templater_Templates](https://github.com/chhoumann/Templater_Templates)
|
||||
- @zachatoo Templates showcase: [https://zachyoung.dev/posts/templater-snippets](https://zachyoung.dev/posts/templater-snippets)
|
||||
- @lguenth Templates showcase: [https://github.com/lguenth/obsidian-templates](https://github.com/lguenth/obsidian-templates)
|
||||
- @tallguyjenks video: [https://youtu.be/2234DXKbNgM?t=1944](https://youtu.be/2234DXKbNgM?t=1944)
|
||||
- @ProductivityGuru videos: [https://www.youtube.com/watch?v=cSawi0tYPMM](https://www.youtube.com/watch?v=cSawi0tYPMM)
|
||||
|
||||
## Alternatives
|
||||
|
||||
- [https://github.com/chhoumann/quickadd](https://github.com/chhoumann/quickadd)
|
||||
- [https://github.com/garyng/obsidian-temple](https://github.com/garyng/obsidian-temple)
|
||||
- [https://github.com/avirut/obsidian-metatemplates](https://github.com/avirut/obsidian-metatemplates)
|
||||
|
||||
## Contributing
|
||||
|
||||
Feel free to contribute.
|
||||
|
||||
You can create an [issue](https://github.com/SilentVoid13/Templater/issues) to report a bug, suggest an improvement for this plugin, ask a question, etc.
|
||||
|
||||
You can make a [pull request](https://github.com/SilentVoid13/Templater/pulls) to contribute to this plugin development.
|
||||
|
||||
Check [this](https://silentvoid13.github.io/Templater/internal-functions/contribute.html) to get more information on how to develop a new internal variable / function.
|
||||
|
||||
## License
|
||||
|
||||
[Templater](https://github.com/SilentVoid13/Templater) is licensed under the GNU AGPLv3 license. Refer to [LICENSE](https://github.com/SilentVoid13/Templater/blob/master/LICENSE.TXT) for more information.
|
||||
|
||||
## Support
|
||||
|
||||
If you want to support me and my work, you can [sponsor me on Github](https://github.com/sponsors/SilentVoid13) (preferred method) or donate something on [**Paypal**](https://www.paypal.com/donate?hosted_button_id=U2SRGAFYXT32Q).
|
||||
33
wiki/concepts/AI配音.md
Normal file
33
wiki/concepts/AI配音.md
Normal file
@@ -0,0 +1,33 @@
|
||||
---
|
||||
id: ai-voice
|
||||
title: "AI配音"
|
||||
type: concept
|
||||
tags: [TTS, voice, audio]
|
||||
sources:
|
||||
- "[[AI配音与声音克隆工具合集]]"
|
||||
last_updated: 2025-03-06
|
||||
---
|
||||
|
||||
## Definition
|
||||
|
||||
AI配音是文本转语音(TTS)技术,将文字内容转化为自然语音。
|
||||
|
||||
## Key Technologies
|
||||
|
||||
- **TTS**:Text-to-Speech,文字转语音
|
||||
- **声音克隆**:用少量样本重建个人声音
|
||||
|
||||
## Popular Tools
|
||||
|
||||
| 平台 | 特点 | 价格 |
|
||||
|------|------|------|
|
||||
| ElevenLabs | 国际顶流,30+语言,情感变化 | 付费较贵 |
|
||||
| 海螺AI | 小白友好,30秒克隆,中文好 | 免费 |
|
||||
| F5-TTS | 开源免费,2秒克隆,技术流 | 免费 |
|
||||
| TTSMaker | 每周3万字,50+语言,300+音色 | 免费限额 |
|
||||
| 剪映 | 抖音官方,短视频首选 | 部分VIP |
|
||||
| AnyVoice | 3秒克隆,中英日韩 | 免费无限 |
|
||||
|
||||
## Connections
|
||||
- [[二创视频]] ← uses ← [[AI配音]]
|
||||
- [[内容创作]] ← uses ← [[AI配音]]
|
||||
42
wiki/concepts/Agent.md
Normal file
42
wiki/concepts/Agent.md
Normal file
@@ -0,0 +1,42 @@
|
||||
---
|
||||
id: agent
|
||||
title: "Agent"
|
||||
type: concept
|
||||
tags: [AI, autonomous, tool-use]
|
||||
sources:
|
||||
- "[[LLM Terms Framework]]"
|
||||
last_updated: 2025-12-20
|
||||
---
|
||||
|
||||
## Definition
|
||||
|
||||
Agent(智能体)是LLM+MCP的组合,LLM负责给出步骤,MCP负责实际执行。
|
||||
|
||||
## How It Works
|
||||
|
||||
1. LLM理解用户意图
|
||||
2. LLM规划执行步骤
|
||||
3. MCP调用外部工具执行
|
||||
4. 结果反馈给LLM
|
||||
5. LLM继续下一步或返回结果
|
||||
|
||||
## Key Capabilities
|
||||
|
||||
- 自主决策
|
||||
- 工具调用
|
||||
- 任务分解
|
||||
- 迭代优化
|
||||
|
||||
## vs Vanilla LLM
|
||||
|
||||
| 维度 | Vanilla LLM | Agent |
|
||||
|------|-------------|-------|
|
||||
| 能力 | 仅生成文本 | 执行实际操作 |
|
||||
| 工具调用 | 无 | 有 |
|
||||
| 自主性 | 低 | 高 |
|
||||
| 幻觉风险 | 高 | 低(可验证) |
|
||||
|
||||
## Connections
|
||||
- [[Agent]] ← combines ← [[LLM]] + [[MCP]]
|
||||
- [[Agent]] ← extends ← [[LLM]]
|
||||
- [[Agent]] ← uses ← [[工具调用]]
|
||||
40
wiki/concepts/AgenticAI.md
Normal file
40
wiki/concepts/AgenticAI.md
Normal file
@@ -0,0 +1,40 @@
|
||||
---
|
||||
id: agentic-ai
|
||||
title: "Agentic AI"
|
||||
type: concept
|
||||
tags: [AI, agent, autonomous, proactive]
|
||||
sources:
|
||||
- "[[Designing for Agentic AI]]"
|
||||
last_updated: 2025-03-02
|
||||
---
|
||||
|
||||
## Definition
|
||||
|
||||
Agentic AI是能够自主行动和决策的AI系统,能够预判用户需求并主动执行任务。
|
||||
|
||||
## Key Characteristics
|
||||
|
||||
- **主动预判**:不需要用户明确指令,主动分析并行动
|
||||
- **实时反馈**:持续向用户展示决策过程
|
||||
- **用户控制**:确保用户对AI行为有最终决定权
|
||||
- **行动执行**:不仅生成内容,而是执行具体操作
|
||||
|
||||
## Five Design Principles
|
||||
|
||||
1. **透明性**:让用户理解AI的决策过程
|
||||
2. **控制权**:用户始终保持对AI行为的最终决定权
|
||||
3. **个性化**:AI适应用户的偏好和习惯
|
||||
4. **对话**:通过自然语言进行持续交互
|
||||
5. **预判**:AI主动识别并满足用户潜在需求
|
||||
|
||||
## vs GenAI
|
||||
|
||||
| 维度 | GenAI | Agentic AI |
|
||||
|------|-------|------------|
|
||||
| 核心能力 | 内容生成 | 行动执行 |
|
||||
| 交互模式 | 被动响应 | 主动预判 |
|
||||
| 反馈机制 | 单次响应 | 实时反馈 |
|
||||
|
||||
## Connections
|
||||
- [[Agentic AI]] ← extends ← [[GenAI]]
|
||||
- [[AI产品设计]] ← uses ← [[Agentic AI设计原则]]
|
||||
55
wiki/concepts/ClaudeSkills.md
Normal file
55
wiki/concepts/ClaudeSkills.md
Normal file
@@ -0,0 +1,55 @@
|
||||
---
|
||||
id: claude-skills
|
||||
title: "Claude Skills"
|
||||
type: concept
|
||||
tags: [Anthropic, Claude, skill, SOP]
|
||||
sources:
|
||||
- "[[Claude Skills最值得研究的AI范式]]"
|
||||
last_updated: 2026-01-05
|
||||
---
|
||||
|
||||
## Definition
|
||||
|
||||
Claude Skills是Anthropic官方发布的AI技能指南,本质是写给Claude的"说明书"和"SOP"。
|
||||
|
||||
## What It Contains
|
||||
|
||||
- Prompt结构定义
|
||||
- 参数含义说明
|
||||
- 容错策略
|
||||
- 使用示例
|
||||
|
||||
## Official Skills Categories
|
||||
|
||||
- 办公自动化四大件(Word/PDF/PPT/Excel)
|
||||
- 开发者工具箱
|
||||
- 创意类Skill
|
||||
|
||||
## Awesome Claude Skills
|
||||
|
||||
三大社区仓库:
|
||||
- ComposioHQ
|
||||
- VoltAgent
|
||||
- BehiSecc
|
||||
|
||||
## Skills聚合站
|
||||
|
||||
- skillsmp.com
|
||||
- aitmpl.com/skills
|
||||
- claudemarketplaces.com
|
||||
|
||||
## Significance
|
||||
|
||||
Skills的爆发标志着从**提示词工程**到**流程工程**的关键转变:
|
||||
- 将经验沉淀为SOP
|
||||
- 交给AI稳定执行
|
||||
- 实现可复用的工作流
|
||||
|
||||
## Connection to Vibe Coding
|
||||
|
||||
Vibe Coding的尽头也是Skills,通过AI编程方式构建的流程最终需要Skills来标准化和复用。
|
||||
|
||||
## Connections
|
||||
- [[提示词工程]] ← evolves_to ← [[流程工程]]
|
||||
- [[Claude Skills]] ← implements ← [[SOP标准化]]
|
||||
- [[Vibe Coding]] ← uses ← [[Claude Skills]]
|
||||
32
wiki/concepts/Embedding.md
Normal file
32
wiki/concepts/Embedding.md
Normal file
@@ -0,0 +1,32 @@
|
||||
---
|
||||
id: embedding
|
||||
title: "Embedding"
|
||||
type: concept
|
||||
tags: [LLM, vector, representation]
|
||||
sources:
|
||||
- "[[RAG从入门到精通系列1:基础RAG]]"
|
||||
- "[[LLM Terms Framework]]"
|
||||
last_updated: 2025-12-18
|
||||
---
|
||||
|
||||
## Definition
|
||||
|
||||
Embedding(向量化)是将文本转换为数值向量的技术,使计算机能够计算词与词之间的距离和语义关系。
|
||||
|
||||
## Mechanism
|
||||
|
||||
- 将文本映射到高维向量空间
|
||||
- 语义相似的文本在向量空间中距离更近
|
||||
- 支持相似度搜索和聚类分析
|
||||
|
||||
## Use Cases
|
||||
|
||||
- RAG系统的文档索引
|
||||
- 语义搜索
|
||||
- 文本相似度比较
|
||||
- 推荐系统
|
||||
|
||||
## Connections
|
||||
- [[LLM]] ← uses ← [[Embedding]]
|
||||
- [[RAG]] ← uses ← [[Embedding]]
|
||||
- [[向量数据库]] ← stores ← [[Embedding]]
|
||||
31
wiki/concepts/GenAI.md
Normal file
31
wiki/concepts/GenAI.md
Normal file
@@ -0,0 +1,31 @@
|
||||
---
|
||||
id: genai
|
||||
title: "GenAI"
|
||||
type: concept
|
||||
tags: [AI, generation, content-creation]
|
||||
sources:
|
||||
- "[[Designing for Agentic AI]]"
|
||||
last_updated: 2025-03-02
|
||||
---
|
||||
|
||||
## Definition
|
||||
|
||||
GenAI(生成式AI)擅长创作内容,如文本、图像、代码、音乐等。
|
||||
|
||||
## Key Characteristics
|
||||
|
||||
- 内容生成能力强
|
||||
- 被动响应用户请求
|
||||
- 适合创意类任务
|
||||
|
||||
## vs Agentic AI
|
||||
|
||||
| 维度 | GenAI | Agentic AI |
|
||||
|------|-------|------------|
|
||||
| 核心能力 | 内容生成 | 行动执行 |
|
||||
| 交互模式 | 被动响应 | 主动预判 |
|
||||
| 代表任务 | 写作、绘画 | 自动化工作流 |
|
||||
|
||||
## Connections
|
||||
- [[Agentic AI]] ← evolves_from ← [[GenAI]]
|
||||
- [[AI产品设计]] ← uses ← [[GenAI]]
|
||||
41
wiki/concepts/LLM.md
Normal file
41
wiki/concepts/LLM.md
Normal file
@@ -0,0 +1,41 @@
|
||||
---
|
||||
id: llm
|
||||
title: "LLM"
|
||||
type: concept
|
||||
tags: [AI, language-model, foundation-model]
|
||||
sources:
|
||||
- "[[LLM Terms Framework]]"
|
||||
last_updated: 2025-12-20
|
||||
---
|
||||
|
||||
## Definition
|
||||
|
||||
LLM(Large Language Model,大语言模型)是参数规模≥1B的深度学习模型,能够理解和生成人类语言。
|
||||
|
||||
## Core Properties
|
||||
|
||||
- **参数规模**:通常≥10亿参数
|
||||
- **语言理解**:能够理解复杂语义
|
||||
- **文本生成**:能够生成连贯、合法的文本
|
||||
- **上下文学习**:能从少量示例中学习
|
||||
|
||||
## Key Metrics
|
||||
|
||||
- **Token**:基本输入单元
|
||||
- 1英文字符 ≈ 0.3 token
|
||||
- 1中文字符 ≈ 0.6 token
|
||||
- **Context Window**:模型能接受的上下文长度
|
||||
|
||||
## Related Concepts
|
||||
|
||||
- [[Token]]:LLM的基本输入单元
|
||||
- [[MCP]]:LLM与外部工具的连接协议
|
||||
- [[Agent]]:LLM+MCP的智能体
|
||||
- [[RAG]]:扩展LLM能力的技术
|
||||
- [[Embedding]]:LLM理解文本的基础
|
||||
|
||||
## Connections
|
||||
- [[LLM]] ← uses ← [[Token]]
|
||||
- [[LLM]] ← uses ← [[MCP]]
|
||||
- [[Agent]] ← combines ← [[LLM]] + [[MCP]]
|
||||
- [[RAG]] ← extends ← [[LLM]]
|
||||
43
wiki/concepts/MCP.md
Normal file
43
wiki/concepts/MCP.md
Normal file
@@ -0,0 +1,43 @@
|
||||
---
|
||||
id: mcp
|
||||
title: "MCP"
|
||||
type: concept
|
||||
tags: [AI, protocol, tool-integration]
|
||||
sources:
|
||||
- "[[LLM Terms Framework]]"
|
||||
last_updated: 2025-12-20
|
||||
---
|
||||
|
||||
## Definition
|
||||
|
||||
MCP(Model Context Protocol,模型上下文协议)是一种标准化接口,用于连接大模型与外部数据和工具。
|
||||
|
||||
## Purpose
|
||||
|
||||
解决LLM无法访问实时数据和外部工具的问题:
|
||||
- LLM给出执行步骤
|
||||
- 实际执行需要配合MCP
|
||||
- 实现智能体(Agent)功能
|
||||
|
||||
## Architecture
|
||||
|
||||
- **Client**:运行在AI应用端
|
||||
- **Server**:运行在外部服务或本地
|
||||
|
||||
## Use Cases
|
||||
|
||||
- 文件系统访问
|
||||
- API调用
|
||||
- 数据库查询
|
||||
- 代码执行
|
||||
|
||||
## Connection to Agent
|
||||
|
||||
Agent = LLM + MCP
|
||||
- LLM负责理解和规划
|
||||
- MCP负责执行具体操作
|
||||
|
||||
## Connections
|
||||
- [[LLM]] ← uses ← [[MCP]]
|
||||
- [[Agent]] ← combines ← [[LLM]] + [[MCP]]
|
||||
- [[MCP]] ← enables ← [[工具调用]]
|
||||
39
wiki/concepts/Prompt能力.md
Normal file
39
wiki/concepts/Prompt能力.md
Normal file
@@ -0,0 +1,39 @@
|
||||
---
|
||||
id: prompt-ability
|
||||
title: "Prompt能力"
|
||||
type: concept
|
||||
tags: [prompt-engineering, communication]
|
||||
sources:
|
||||
- "[[如何写出完美的Prompt]]"
|
||||
last_updated: 2025-12-02
|
||||
---
|
||||
|
||||
## Definition
|
||||
|
||||
Prompt能力是清晰界定需求+结构化思维表达的能力,本质是需求拆解+结构化表达能力。
|
||||
|
||||
## Core Elements
|
||||
|
||||
人与AI的协作协议,定义:
|
||||
- **做什么**:明确任务目标
|
||||
- **为什么**:任务背景和目的
|
||||
- **给谁**:目标受众
|
||||
- **怎么做**:执行方式和约束
|
||||
- **做到什么标准**:质量要求和验收标准
|
||||
|
||||
## Four Key Elements
|
||||
|
||||
1. **角色**:AI扮演的身份
|
||||
2. **受众对齐**:明确目标用户
|
||||
3. **场景对齐**:使用环境上下文
|
||||
4. **目标对齐**:预期成果定义
|
||||
|
||||
## Common Mistakes
|
||||
|
||||
- 越复杂越专业
|
||||
- 说清做什么就行
|
||||
- 一键生成即终点
|
||||
|
||||
## Connections
|
||||
- [[AI协作]] ← requires ← [[Prompt能力]]
|
||||
- [[结构化思维]] ← enables ← [[Prompt能力]]
|
||||
41
wiki/concepts/RAG.md
Normal file
41
wiki/concepts/RAG.md
Normal file
@@ -0,0 +1,41 @@
|
||||
---
|
||||
id: rag
|
||||
title: "RAG"
|
||||
type: concept
|
||||
tags: [LLM, retrieval, augmentation]
|
||||
sources:
|
||||
- "[[RAG从入门到精通系列1:基础RAG]]"
|
||||
- "[[LLM Terms Framework]]"
|
||||
last_updated: 2025-12-18
|
||||
---
|
||||
|
||||
## Definition
|
||||
|
||||
RAG(Retrieval Augmented Generation,检索增强生成)是一种结合检索系统和LLM生成的技术,解决LLM缺乏最新和私有数据的问题。
|
||||
|
||||
## Three-Step Process
|
||||
|
||||
1. **索引(Indexing)**:将文档切分并转换为Embedding向量存入向量数据库
|
||||
2. **检索(Retrieval)**:根据问题语义向量检索相关文档块
|
||||
3. **生成(Generation)**:将问题和相关文档输入LLM生成答案
|
||||
|
||||
## Key Components
|
||||
|
||||
- **Embedding**:将文本转换为数值向量
|
||||
- **向量数据库**:存储和检索向量表示(如Qdrant)
|
||||
- **文档切分**:将长文档分割成符合Embedding窗口的块
|
||||
- **Context Window**:模型能接受的上下文长度限制(512-8192 token)
|
||||
|
||||
## Why It Matters
|
||||
|
||||
解决LLM的幻觉问题,让模型能够:
|
||||
- 访问最新信息
|
||||
- 利用私有数据
|
||||
- 提供可溯源的回答
|
||||
|
||||
## Connections
|
||||
- [[LLM]] ← uses ← [[RAG]]
|
||||
- [[RAG]] ← includes ← [[索引]]
|
||||
- [[RAG]] ← includes ← [[检索]]
|
||||
- [[RAG]] ← includes ← [[生成]]
|
||||
- [[RAG]] ← extends ← [[LLM]]
|
||||
30
wiki/concepts/SourceGrounding.md
Normal file
30
wiki/concepts/SourceGrounding.md
Normal file
@@ -0,0 +1,30 @@
|
||||
---
|
||||
id: source-grounding
|
||||
title: "Source-Grounding"
|
||||
type: concept
|
||||
tags: [NotebookLM, accuracy, grounding]
|
||||
sources:
|
||||
- "[[7 ways I use NotebookLM to make my life easier]]"
|
||||
last_updated: 2025-11-23
|
||||
---
|
||||
|
||||
## Definition
|
||||
|
||||
Source-Grounding是NotebookLM的核心机制,限制知识库仅包含用户上传的文档,确保AI回答准确且可溯源。
|
||||
|
||||
## Mechanism
|
||||
|
||||
- 用户上传文档后,NotebookLM只在这个文档范围内回答
|
||||
- 避免AI幻觉,确保回答有据可查
|
||||
- 每个回答都附带源文档引用
|
||||
|
||||
## Why It Matters
|
||||
|
||||
解决通用LLM的幻觉问题,特别适用于:
|
||||
- 法律文档审查
|
||||
- 学术研究
|
||||
- 精确信息查询
|
||||
|
||||
## Connections
|
||||
- [[NotebookLM]] ← uses ← [[Source-Grounding]]
|
||||
- [[AI准确性]] ← requires ← [[Source-Grounding]]
|
||||
37
wiki/concepts/Token.md
Normal file
37
wiki/concepts/Token.md
Normal file
@@ -0,0 +1,37 @@
|
||||
---
|
||||
id: token
|
||||
title: "Token"
|
||||
type: concept
|
||||
tags: [LLM, tokenization, input-unit]
|
||||
sources:
|
||||
- "[[LLM Terms Framework]]"
|
||||
last_updated: 2025-12-20
|
||||
---
|
||||
|
||||
## Definition
|
||||
|
||||
Token是大模型的基本输入单元,是文本处理的最小单位。
|
||||
|
||||
## Tokenization Rules
|
||||
|
||||
- 1英文字符 ≈ 0.3 token
|
||||
- 1中文字符 ≈ 0.6 token
|
||||
- 标点符号和空格也占用token
|
||||
|
||||
## Why It Matters
|
||||
|
||||
- 影响API调用成本
|
||||
- 决定上下文长度限制
|
||||
- 影响生成速度
|
||||
|
||||
## Context Window
|
||||
|
||||
模型能接受的token数量限制:
|
||||
- 较短的模型:4K-8K tokens
|
||||
- 中等模型:32K-128K tokens
|
||||
- 长上下文模型:1M+ tokens
|
||||
|
||||
## Connections
|
||||
- [[LLM]] ← uses ← [[Token]]
|
||||
- [[Token]] → affects → [[成本计算]]
|
||||
- [[Token]] → affects → [[上下文限制]]
|
||||
32
wiki/concepts/VibeCoding.md
Normal file
32
wiki/concepts/VibeCoding.md
Normal file
@@ -0,0 +1,32 @@
|
||||
---
|
||||
id: vibe-coding
|
||||
title: "Vibe Coding"
|
||||
type: concept
|
||||
tags: [AI, programming, coding]
|
||||
sources:
|
||||
- "[[Claude Skills最值得研究的AI范式]]"
|
||||
last_updated: 2026-01-05
|
||||
---
|
||||
|
||||
## Definition
|
||||
|
||||
Vibe Coding是一种AI编程方式,通过自然语言与AI协作编写代码。
|
||||
|
||||
## Characteristics
|
||||
|
||||
- 自然语言为主
|
||||
- AI生成代码
|
||||
- 人类审核和调整
|
||||
- 降低编程门槛
|
||||
|
||||
## The End State
|
||||
|
||||
Vibe Coding的尽头是Skills:
|
||||
- 通过对话构建的代码和流程
|
||||
- 需要标准化为Skills以便复用
|
||||
- 最终沉淀为可维护的系统
|
||||
|
||||
## Connections
|
||||
- [[Vibe Coding]] ← uses ← [[Claude Skills]]
|
||||
- [[AI编程]] ← extends ← [[Vibe Coding]]
|
||||
- [[提示词工程]] ← relates_to ← [[Vibe Coding]]
|
||||
66
wiki/concepts/Workspace.md
Normal file
66
wiki/concepts/Workspace.md
Normal file
@@ -0,0 +1,66 @@
|
||||
---
|
||||
title: "Workspace"
|
||||
type: concept
|
||||
tags: [openclaw, workspace, configuration]
|
||||
last_updated: 2026-04-14
|
||||
---
|
||||
|
||||
# Workspace
|
||||
|
||||
OpenClaw中Agent的工作台目录,决定Agent如何工作。
|
||||
|
||||
## 核心文件
|
||||
|
||||
### AGENTS.md
|
||||
定义Agent的:
|
||||
- 岗位职责
|
||||
- 行为边界
|
||||
- 多Agent协调规则
|
||||
|
||||
### SOUL.md
|
||||
定义Agent的:
|
||||
- 性格叙事
|
||||
- 沟通风格
|
||||
- 价值观和边界
|
||||
|
||||
### USER.md
|
||||
固化用户的:
|
||||
- 偏好设定
|
||||
- 背景知识假设
|
||||
- 常见任务
|
||||
|
||||
### TOOLS.md
|
||||
声明工具的:
|
||||
- 可用工具
|
||||
- 使用原则
|
||||
- 受限工具
|
||||
|
||||
### IDENTITY.md
|
||||
结构化身份档案:
|
||||
- 名字
|
||||
- 角色类型
|
||||
- Emoji
|
||||
- 头像
|
||||
|
||||
### BOOTSTRAP.md
|
||||
一次性启动引导,完成后应删除。
|
||||
|
||||
### memory/
|
||||
长期记忆目录:
|
||||
- 按日期滚动的记忆笔记
|
||||
- 实现跨会话上下文保留
|
||||
|
||||
## 配置要点
|
||||
|
||||
1. **边界比能力更重要**:明确"不要做什么"
|
||||
2. **场景触发优于通用指令**:具体场景下的具体规则
|
||||
3. **简洁有效**:300-500字的AGENTS.md比2000字的更有效
|
||||
|
||||
## 与openclaw.json的关系
|
||||
- Workspace文件:管"Agent平时怎么干活"
|
||||
- openclaw.json:管"系统怎么跑Agent"
|
||||
|
||||
## 相关概念
|
||||
- [[OpenClaw]]
|
||||
- [[AGENTS.md]]
|
||||
- [[SOUL.md]]
|
||||
36
wiki/concepts/vLLM.md
Normal file
36
wiki/concepts/vLLM.md
Normal file
@@ -0,0 +1,36 @@
|
||||
---
|
||||
id: vllm
|
||||
title: "vLLM"
|
||||
type: concept
|
||||
tags: [LLM, inference, GPU, optimization]
|
||||
sources:
|
||||
- "[[LLM Terms Framework]]"
|
||||
last_updated: 2025-12-20
|
||||
---
|
||||
|
||||
## Definition
|
||||
|
||||
vLLM是一个高效LLM推理框架,通过KV Cache和连续批处理提升GPU利用率。
|
||||
|
||||
## Key Optimizations
|
||||
|
||||
### KV Cache
|
||||
- 缓存已计算的Key-Value矩阵
|
||||
- 避免重复计算
|
||||
- 大幅提升推理速度
|
||||
|
||||
### Continuous Batching
|
||||
- 动态批处理多个请求
|
||||
- 提高GPU利用率
|
||||
- 降低延迟
|
||||
|
||||
## Why It Matters
|
||||
|
||||
- 官方HuggingFace推理速度慢
|
||||
- vLLM可提升10-24倍速度
|
||||
- 支持高并发推理
|
||||
|
||||
## Connections
|
||||
- [[LLM]] ← uses ← [[vLLM]]
|
||||
- [[推理优化]] ← uses ← [[vLLM]]
|
||||
- [[GPU利用率]] ← improves ← [[vLLM]]
|
||||
38
wiki/concepts/九宫格法.md
Normal file
38
wiki/concepts/九宫格法.md
Normal file
@@ -0,0 +1,38 @@
|
||||
---
|
||||
id: nine-grid
|
||||
title: "九宫格法"
|
||||
type: concept
|
||||
tags: [video, AI, image-generation]
|
||||
sources:
|
||||
- "[[固定镜头短视频AI全流程制作]]"
|
||||
last_updated: 2025-03-15
|
||||
---
|
||||
|
||||
## Definition
|
||||
|
||||
九宫格法是一次性生成3x3共九个分镜画面的方法,确保多个镜头之间的画面一致性。
|
||||
|
||||
## Mechanism
|
||||
|
||||
1. 将视频分割为9个分镜
|
||||
2. 一次性生成3x3网格图像
|
||||
3. 每个格子是一个分镜的关键帧
|
||||
4. 确保人物/场景在多个格子中保持一致
|
||||
|
||||
## Why It Works
|
||||
|
||||
- AI在单张图像内保持一致性更容易
|
||||
- 避免逐帧生成导致的人物变形
|
||||
- 提高多镜头视频的整体质量
|
||||
|
||||
## Five-Step Formula
|
||||
|
||||
1. 拆分镜头
|
||||
2. 一致性图像生成(九宫格法)
|
||||
3. 首尾针动画
|
||||
4. 快速剪辑
|
||||
5. 声音设计
|
||||
|
||||
## Connections
|
||||
- [[AI视频制作]] ← uses ← [[九宫格法]]
|
||||
- [[分镜设计]] ← uses ← [[九宫格法]]
|
||||
37
wiki/concepts/固定机位.md
Normal file
37
wiki/concepts/固定机位.md
Normal file
@@ -0,0 +1,37 @@
|
||||
---
|
||||
id: fixed-camera
|
||||
title: "固定机位"
|
||||
type: concept
|
||||
tags: [video-production, cinematography]
|
||||
sources:
|
||||
- "[[固定镜头短视频AI全流程制作]]"
|
||||
last_updated: 2025-03-15
|
||||
---
|
||||
|
||||
## Definition
|
||||
|
||||
固定机位是摄像机位置固定不变的拍摄方式,是固定镜头短视频的核心特征。
|
||||
|
||||
## Key Characteristics
|
||||
|
||||
- 摄像机位置不变
|
||||
- 只有画面内容变化
|
||||
- 适合展示时间流逝
|
||||
- 便于AI生成一致性画面
|
||||
|
||||
## Use Cases
|
||||
|
||||
- 家装视频
|
||||
- 产品展示
|
||||
- 教程演示
|
||||
- 时间压缩视频
|
||||
|
||||
## Connection to AI Video
|
||||
|
||||
固定机位降低AI视频生成的复杂度,通过:
|
||||
- 九宫格法保证画面一致性
|
||||
- 首尾针动画实现平滑过渡
|
||||
|
||||
## Connections
|
||||
- [[AI视频制作]] ← uses ← [[固定机位]]
|
||||
- [[短视频制作]] ← uses ← [[固定机位]]
|
||||
32
wiki/concepts/声音克隆.md
Normal file
32
wiki/concepts/声音克隆.md
Normal file
@@ -0,0 +1,32 @@
|
||||
---
|
||||
id: voice-cloning
|
||||
title: "声音克隆"
|
||||
type: concept
|
||||
tags: [TTS, voice, cloning]
|
||||
sources:
|
||||
- "[[AI配音与声音克隆工具合集]]"
|
||||
last_updated: 2025-03-06
|
||||
---
|
||||
|
||||
## Definition
|
||||
|
||||
声音克隆是用少量音频样本重建个人声音特征的技术。
|
||||
|
||||
## How It Works
|
||||
|
||||
1. 收集目标声音的短音频(2-30秒)
|
||||
2. 提取声音特征
|
||||
3. 生成新的语音内容
|
||||
|
||||
## Speed Comparison
|
||||
|
||||
| 工具 | 克隆速度 | 技术门槛 |
|
||||
|------|----------|----------|
|
||||
| F5-TTS | 2秒 | 高(需代码) |
|
||||
| 海螺AI | 30秒 | 低 |
|
||||
| AnyVoice | 3秒 | 低 |
|
||||
| ElevenLabs | 30秒 | 低 |
|
||||
|
||||
## Connections
|
||||
- [[AI配音]] ← uses ← [[声音克隆]]
|
||||
- [[内容创作]] ← uses ← [[声音克隆]]
|
||||
64
wiki/concepts/多Agent系统.md
Normal file
64
wiki/concepts/多Agent系统.md
Normal file
@@ -0,0 +1,64 @@
|
||||
---
|
||||
title: "多Agent系统"
|
||||
type: concept
|
||||
tags: [multi-agent, collaboration, agent]
|
||||
last_updated: 2026-04-14
|
||||
---
|
||||
|
||||
# 多Agent系统
|
||||
|
||||
多个专业Agent协同工作的架构模式,每个Agent有独特的角色和职责。
|
||||
|
||||
## 核心模式
|
||||
|
||||
### 分散式协调
|
||||
通过共享STATE.yaml文件协调,而非中央orchestrator:
|
||||
- Agent读写共享状态文件
|
||||
- 多子Agent并行工作
|
||||
- 主会话保持精简(CEO模式)
|
||||
|
||||
### STATE.yaml
|
||||
项目协调文件,作为单一事实来源:
|
||||
```yaml
|
||||
project: website-redesign
|
||||
tasks:
|
||||
- id: homepage-hero
|
||||
status: in_progress
|
||||
owner: pm-frontend
|
||||
```
|
||||
|
||||
### 团队配置示例
|
||||
- [[Milo]]:策略Lead
|
||||
- [[Josh]]:商业分析
|
||||
- Marketing Agent:营销研究
|
||||
- Dev Agent:开发
|
||||
|
||||
## 关键优势
|
||||
|
||||
1. **专业化分工**:每个Agent专注特定领域
|
||||
2. **并行执行**:多任务同时处理
|
||||
3. **可扩展性**:新增Agent无需修改主逻辑
|
||||
4. **共享记忆**:团队成员共享项目上下文
|
||||
|
||||
## 协作机制
|
||||
|
||||
- **Telegram路由**:通过标签分配到不同Agent
|
||||
- **共享内存**:项目文档、目标、决策
|
||||
- **私有上下文**:每个Agent独有会话历史
|
||||
- **定时任务**:Agent主动工作
|
||||
|
||||
## Race Condition处理
|
||||
|
||||
当多个Agent编辑同一文件时:
|
||||
1. AUTONOMOUS.md:仅主会话编辑
|
||||
2. memory/tasks-log.md:仅追加,子Agent只添加新行
|
||||
|
||||
## 使用场景
|
||||
|
||||
- [[多Agent专业团队]]
|
||||
- [[多Agent内容工厂]]
|
||||
- [[自主项目管理]]
|
||||
- [[动态仪表板]]
|
||||
|
||||
## 相关链接
|
||||
- [Anthropic: Building Effective Agents](https://www.anthropic.com/research/building-effective-agents)
|
||||
59
wiki/concepts/工作流自动化.md
Normal file
59
wiki/concepts/工作流自动化.md
Normal file
@@ -0,0 +1,59 @@
|
||||
---
|
||||
title: "工作流自动化"
|
||||
type: concept
|
||||
tags: [automation, workflow, n8n]
|
||||
last_updated: 2026-04-14
|
||||
---
|
||||
|
||||
# 工作流自动化
|
||||
|
||||
使用工具自动执行重复性任务,减少人工干预。
|
||||
|
||||
## 核心概念
|
||||
|
||||
### 工作流(Workflow)
|
||||
由多个任务节点按一定顺序执行的自动化流程。
|
||||
|
||||
### 节点(Node)
|
||||
工作流中的单个操作单元:
|
||||
- 触发器:启动工作流
|
||||
- 动作:执行具体操作
|
||||
- 工具:辅助功能
|
||||
- 代码:自定义逻辑
|
||||
- AI节点:嵌入AI能力
|
||||
|
||||
## 与AI Agent的关系
|
||||
|
||||
- **Workflow**:预定义自动化,一致输出
|
||||
- **Agent**:基于LLM动态决定工具和输出
|
||||
- **Agentic系统**:结合两者优势
|
||||
|
||||
## 平台
|
||||
|
||||
### N8N
|
||||
- 可视化拖拽界面
|
||||
- 400+预构建集成
|
||||
- 支持自托管
|
||||
|
||||
### OpenClaw
|
||||
- 通过skill扩展能力
|
||||
- 自然语言配置
|
||||
- 记忆和上下文保留
|
||||
|
||||
## 安全集成模式
|
||||
|
||||
[[OpenClaw + n8n工作流编排]]:
|
||||
- Webhook调用n8n
|
||||
- 凭证隔离在n8n
|
||||
- 工作流可锁定
|
||||
|
||||
## 使用场景
|
||||
|
||||
- [[会议纪要自动化]]
|
||||
- [[邮件管理自动化]]
|
||||
- [[日历聚合]]
|
||||
- [[社交媒体自动化]]
|
||||
|
||||
## 相关链接
|
||||
- [N8N官网](https://n8n.io/)
|
||||
- [OpenClaw文档](https://docs.openclaw.ai)
|
||||
31
wiki/concepts/思维链引导.md
Normal file
31
wiki/concepts/思维链引导.md
Normal file
@@ -0,0 +1,31 @@
|
||||
---
|
||||
id: chain-of-thought
|
||||
title: "思维链引导"
|
||||
type: concept
|
||||
tags: [prompt-engineering, reasoning]
|
||||
sources:
|
||||
- "[[如何写出完美的Prompt]]"
|
||||
last_updated: 2025-12-02
|
||||
---
|
||||
|
||||
## Definition
|
||||
|
||||
思维链引导是一种提示词技术,让AI逐步推理而非直接给出答案。
|
||||
|
||||
## Mechanism
|
||||
|
||||
通过在提示词中要求AI展示推理过程:
|
||||
- 先分析问题
|
||||
- 再列出步骤
|
||||
- 最后给出答案
|
||||
|
||||
## Benefits
|
||||
|
||||
- 提高AI推理准确性
|
||||
- 减少幻觉发生
|
||||
- 让用户理解决策过程
|
||||
- 便于发现AI思维漏洞
|
||||
|
||||
## Connections
|
||||
- [[Prompt能力]] ← uses ← [[思维链引导]]
|
||||
- [[需求拆解]] ← extends ← [[思维链引导]]
|
||||
62
wiki/concepts/技能系统.md
Normal file
62
wiki/concepts/技能系统.md
Normal file
@@ -0,0 +1,62 @@
|
||||
---
|
||||
title: "技能系统"
|
||||
type: concept
|
||||
tags: [skill, openclaw, extension]
|
||||
last_updated: 2026-04-14
|
||||
---
|
||||
|
||||
# 技能系统
|
||||
|
||||
OpenClaw的扩展机制,通过技能包添加新能力。
|
||||
|
||||
## 技能结构
|
||||
|
||||
```
|
||||
skills/
|
||||
├── skill-name/
|
||||
│ └── SKILL.md
|
||||
```
|
||||
|
||||
## 常用技能
|
||||
|
||||
### 集成技能
|
||||
- [[Telegram]]:消息通道
|
||||
- [[Discord]]:协作平台
|
||||
- [[Slack]]:团队通讯
|
||||
|
||||
### 数据技能
|
||||
- [[YouTube]]:视频内容获取
|
||||
- [[Reddit]]:社区内容聚合
|
||||
- [[GitHub]]:代码和项目数据
|
||||
|
||||
### 工具技能
|
||||
- [[arxiv-reader]]:学术论文读取
|
||||
- [[latex-compiler]]:LaTeX编译
|
||||
- [[youtube-full]]:YouTube完整集成
|
||||
|
||||
### MCP技能
|
||||
- [[n8n-mcp]]:N8N节点访问
|
||||
- [[idea-reality-mcp]]:创意验证
|
||||
|
||||
## 技能安装
|
||||
|
||||
通过ClawHub安装:
|
||||
```bash
|
||||
npx clawhub@latest install skill-name
|
||||
```
|
||||
|
||||
或通过OpenClaw:
|
||||
```text
|
||||
Install the youtube-full skill
|
||||
```
|
||||
|
||||
## 技能开发
|
||||
|
||||
技能是包含SKILL.md的目录,定义:
|
||||
- 工具列表
|
||||
- 使用方法
|
||||
- 配置要求
|
||||
|
||||
## 使用场景
|
||||
|
||||
详见各use case中的"Skills you Need"部分。
|
||||
33
wiki/concepts/提示词框架.md
Normal file
33
wiki/concepts/提示词框架.md
Normal file
@@ -0,0 +1,33 @@
|
||||
---
|
||||
id: prompt-framework
|
||||
title: "提示词框架"
|
||||
type: concept
|
||||
tags: [Nano Banana, prompt-engineering, image-generation]
|
||||
sources:
|
||||
- "[[Nano Banana提示词框架]]"
|
||||
last_updated: 2025-03-15
|
||||
---
|
||||
|
||||
## Definition
|
||||
|
||||
提示词框架是结构化描述图像生成需求的模板,通过标准化字段确保生成质量可控。
|
||||
|
||||
## Framework Types
|
||||
|
||||
### 物件描述框架
|
||||
- shot:镜头类型
|
||||
- subject:包含item/materials/details/condition
|
||||
- environment:环境描述
|
||||
- lighting:光线
|
||||
- camera:相机设置
|
||||
- color_grade:色彩分级
|
||||
- style:风格
|
||||
- quality:质量参数
|
||||
- negatives:负面提示
|
||||
|
||||
### 人物描述框架
|
||||
- subject:包含age/appearance/pose等字段
|
||||
|
||||
## Connections
|
||||
- [[AI图像生成]] ← uses ← [[提示词框架]]
|
||||
- [[Nano Banana]] ← supports ← [[提示词框架]]
|
||||
34
wiki/concepts/流程工程.md
Normal file
34
wiki/concepts/流程工程.md
Normal file
@@ -0,0 +1,34 @@
|
||||
---
|
||||
id: workflow-engineering
|
||||
title: "流程工程"
|
||||
type: concept
|
||||
tags: [AI, workflow, SOP, engineering]
|
||||
sources:
|
||||
- "[[Claude Skills最值得研究的AI范式]]"
|
||||
last_updated: 2026-01-05
|
||||
---
|
||||
|
||||
## Definition
|
||||
|
||||
流程工程是将重复任务拆解为AI能理解、稳定复用的流程,并通过Skills实现标准化的工程化方法。
|
||||
|
||||
## vs 提示词工程
|
||||
|
||||
| 维度 | 提示词工程 | 流程工程 |
|
||||
|------|------------|----------|
|
||||
| 核心 | 单次Prompt优化 | 全流程标准化 |
|
||||
| 稳定性 | 依赖模型表现 | SOP固化 |
|
||||
| 复用性 | 低 | 高 |
|
||||
| 目标 | 一次好结果 | 稳定可重复 |
|
||||
|
||||
## Key Elements
|
||||
|
||||
- **SOP标准化**:将经验沉淀为操作步骤
|
||||
- **Skills封装**:AI技能的模块化
|
||||
- **自动化执行**:交给AI稳定运行
|
||||
- **反馈迭代**:持续优化流程
|
||||
|
||||
## Connections
|
||||
- [[提示词工程]] ← evolves_to ← [[流程工程]]
|
||||
- [[Claude Skills]] ← implements ← [[流程工程]]
|
||||
- [[SOP标准化]] ← enables ← [[流程工程]]
|
||||
31
wiki/concepts/结构化表达.md
Normal file
31
wiki/concepts/结构化表达.md
Normal file
@@ -0,0 +1,31 @@
|
||||
---
|
||||
id: structured-expression
|
||||
title: "结构化表达"
|
||||
type: concept
|
||||
tags: [prompt-engineering, communication]
|
||||
sources:
|
||||
- "[[如何写出完美的Prompt]]"
|
||||
last_updated: 2025-12-02
|
||||
---
|
||||
|
||||
## Definition
|
||||
|
||||
结构化表达是用清晰逻辑组织信息的方法,确保AI准确理解人类意图。
|
||||
|
||||
## Principles
|
||||
|
||||
- 层次分明:按重要性和逻辑顺序组织
|
||||
- 格式统一:使用一致的标记和分隔符
|
||||
- 信息完整:不遗漏关键上下文
|
||||
- 表达精准:避免歧义和模糊表述
|
||||
|
||||
## Techniques
|
||||
|
||||
- 使用编号列表组织要点
|
||||
- 使用标题区分不同部分
|
||||
- 使用表格呈现结构化数据
|
||||
- 使用引用标记重要信息
|
||||
|
||||
## Connections
|
||||
- [[Prompt能力]] ← enables ← [[结构化表达]]
|
||||
- [[结构化思维]] ← implements ← [[结构化表达]]
|
||||
63
wiki/concepts/记忆系统.md
Normal file
63
wiki/concepts/记忆系统.md
Normal file
@@ -0,0 +1,63 @@
|
||||
---
|
||||
title: "记忆系统"
|
||||
type: concept
|
||||
tags: [memory, openclaw, context]
|
||||
last_updated: 2026-04-14
|
||||
---
|
||||
|
||||
# 记忆系统
|
||||
|
||||
AI Agent跨会话保留上下文和知识的能力。
|
||||
|
||||
## OpenClaw记忆机制
|
||||
|
||||
### 内置方案(builtin)
|
||||
原始记忆存储在Markdown文件中,系统维护本地索引方便检索。
|
||||
|
||||
### QMD方案
|
||||
围绕workspace中的Markdown文件,使用更强的检索/索引方式。
|
||||
|
||||
### 记忆流程
|
||||
```
|
||||
对话发生
|
||||
↓
|
||||
Agent通过普通文件工具把重要信息写入memory/或MEMORY.md
|
||||
↓
|
||||
下次对话开始
|
||||
↓
|
||||
Agent通过memory_search/memory_get检索相关记忆
|
||||
↓
|
||||
相关记忆被注入到当前对话上下文
|
||||
↓
|
||||
Agent表现出"我记得你说过……"的能力
|
||||
```
|
||||
|
||||
## 向量语义搜索
|
||||
|
||||
[[Semantic Memory Search]]使用memsearch:
|
||||
- 索引Markdown记忆文件到向量数据库
|
||||
- 通过含义搜索而非关键词
|
||||
- SHA-256内容哈希避免重复嵌入
|
||||
|
||||
## Workspace记忆文件
|
||||
|
||||
- [[memory/]]:按日期滚动的记忆笔记
|
||||
- [[MEMORY.md]]:长期知识总表
|
||||
- 与memory/目录兼容
|
||||
|
||||
## 关键洞察
|
||||
|
||||
- 对Agent来说,真正算数的长期记忆是Markdown文件
|
||||
- 向量索引只是派生缓存,可以随时重建
|
||||
- 文件永不修改
|
||||
|
||||
## 使用场景
|
||||
|
||||
- [[第二大脑]]
|
||||
- [[个人CRM]]
|
||||
- [[健康症状追踪]]
|
||||
|
||||
## 相关工具
|
||||
|
||||
- [[memsearch]]:向量语义搜索工具
|
||||
- [[Milvus]]:向量数据库后端
|
||||
36
wiki/concepts/需求拆解.md
Normal file
36
wiki/concepts/需求拆解.md
Normal file
@@ -0,0 +1,36 @@
|
||||
---
|
||||
id: requirement-decomposition
|
||||
title: "需求拆解"
|
||||
type: concept
|
||||
tags: [prompt-engineering, structured-thinking]
|
||||
sources:
|
||||
- "[[如何写出完美的Prompt]]"
|
||||
last_updated: 2025-12-02
|
||||
---
|
||||
|
||||
## Definition
|
||||
|
||||
需求拆解是将模糊目标转化为具体可执行子任务的过程。
|
||||
|
||||
## Methods
|
||||
|
||||
### 基础方法
|
||||
- **需求拆解法**:将复杂任务分解为简单步骤
|
||||
- **上下文补全法**:补充背景信息让AI理解场景
|
||||
- **格式定义法**:明确输出格式要求
|
||||
- **示例引导法**:提供参考案例
|
||||
|
||||
### 进阶策略
|
||||
- **思维链引导**:让AI逐步推理
|
||||
- **任务拆分**:大任务分解为子任务
|
||||
- **角色赋能**:赋予AI特定专业角色
|
||||
- **预填回复**:提供初始回答框架
|
||||
|
||||
### 高阶技巧
|
||||
- **跨模态联动**:结合多种输入输出形式
|
||||
- **领域知识注入**:嵌入专业知识
|
||||
- **反馈循环嵌入**:建立迭代优化机制
|
||||
|
||||
## Connections
|
||||
- [[Prompt能力]] ← requires ← [[需求拆解]]
|
||||
- [[结构化表达]] ← enables ← [[需求拆解]]
|
||||
29
wiki/concepts/音频概览.md
Normal file
29
wiki/concepts/音频概览.md
Normal file
@@ -0,0 +1,29 @@
|
||||
---
|
||||
id: audio-overview
|
||||
title: "音频概览"
|
||||
type: concept
|
||||
tags: [NotebookLM, learning, podcast]
|
||||
sources:
|
||||
- "[[7 ways I use NotebookLM to make my life easier]]"
|
||||
last_updated: 2025-11-23
|
||||
---
|
||||
|
||||
## Definition
|
||||
|
||||
音频概览是NotebookLM的功能,将文档转化为AI双人播客格式,适合被动学习。
|
||||
|
||||
## Mechanism
|
||||
|
||||
- AI分析文档内容
|
||||
- 生成两个AI声音的对话
|
||||
- 用户可以收听而非阅读
|
||||
|
||||
## Use Cases
|
||||
|
||||
- 通勤时学习
|
||||
- 视觉疲劳时继续学习
|
||||
- 将长文档转化为可听的摘要
|
||||
|
||||
## Connections
|
||||
- [[NotebookLM]] ← implements ← [[音频概览]]
|
||||
- [[被动学习]] ← uses ← [[音频概览]]
|
||||
37
wiki/concepts/首尾针动画.md
Normal file
37
wiki/concepts/首尾针动画.md
Normal file
@@ -0,0 +1,37 @@
|
||||
---
|
||||
id: frame-interpolation
|
||||
title: "首尾针动画"
|
||||
type: concept
|
||||
tags: [video, AI, animation]
|
||||
sources:
|
||||
- "[[固定镜头短视频AI全流程制作]]"
|
||||
last_updated: 2025-03-15
|
||||
---
|
||||
|
||||
## Definition
|
||||
|
||||
首尾针动画是通过AI自动补齐首尾帧之间中间动作的技术,实现平滑过渡效果。
|
||||
|
||||
## Mechanism
|
||||
|
||||
1. 确定起始帧和结束帧
|
||||
2. AI分析首尾帧的差异
|
||||
3. 自动生成中间过渡帧
|
||||
4. 输出流畅视频
|
||||
|
||||
## Tools
|
||||
|
||||
- 海螺AI
|
||||
- KAI
|
||||
- 其他AI动效工具
|
||||
|
||||
## Connection to Fixed-Camera Video
|
||||
|
||||
固定机位视频天然适合首尾针动画,因为:
|
||||
- 背景固定,减少AI生成负担
|
||||
- 只需关注主体变化
|
||||
- 更容易保持一致性
|
||||
|
||||
## Connections
|
||||
- [[AI视频制作]] ← uses ← [[首尾针动画]]
|
||||
- [[固定机位]] ← enables ← [[首尾针动画]]
|
||||
29
wiki/entities/DeepSider.md
Normal file
29
wiki/entities/DeepSider.md
Normal file
@@ -0,0 +1,29 @@
|
||||
---
|
||||
id: deepsider
|
||||
title: "DeepSider"
|
||||
type: entity
|
||||
tags: [browser-extension, AI, multi-model]
|
||||
sources:
|
||||
- "[[Nano Banana 2使用指南]]"
|
||||
last_updated: 2025-12-01
|
||||
---
|
||||
|
||||
## Overview
|
||||
|
||||
DeepSider是浏览器插件,聚合多AI模型供用户使用,国内可直接访问。
|
||||
|
||||
## Key Features
|
||||
|
||||
- 支持Gemini 3.0、Gemini 3 Pro Image(Nano Banana 2)、GPT-5.1等模型
|
||||
- 国内可用,绕过访问限制
|
||||
- 浏览器插件形式,便捷集成
|
||||
|
||||
## Use Cases
|
||||
|
||||
- 访问Gemini系列模型
|
||||
- AI图像生成(Nano Banana 2)
|
||||
- 多模型对比使用
|
||||
|
||||
## Connections
|
||||
- [[DeepSider]] → enables → [[Nano Banana 2]]
|
||||
- [[AI工具聚合]] ← uses ← [[DeepSider]]
|
||||
28
wiki/entities/ElevenLabs.md
Normal file
28
wiki/entities/ElevenLabs.md
Normal file
@@ -0,0 +1,28 @@
|
||||
---
|
||||
id: elevenlabs
|
||||
title: "ElevenLabs"
|
||||
type: entity
|
||||
tags: [AI, TTS, voice-cloning, international]
|
||||
sources:
|
||||
- "[[AI配音与声音克隆工具合集]]"
|
||||
last_updated: 2025-03-06
|
||||
---
|
||||
|
||||
## Overview
|
||||
|
||||
ElevenLabs是国际顶流AI配音和声音克隆平台,支持30+语言和情感变化,付费较高。
|
||||
|
||||
## Key Features
|
||||
|
||||
- 30+语言支持
|
||||
- 情感变化控制
|
||||
- 高保真声音克隆
|
||||
- 国际顶流品质
|
||||
|
||||
## Pricing
|
||||
|
||||
付费较贵,适合专业用户
|
||||
|
||||
## Connections
|
||||
- [[AI配音]] ← uses ← [[ElevenLabs]]
|
||||
- [[声音克隆]] ← uses ← [[ElevenLabs]]
|
||||
25
wiki/entities/F5TTS.md
Normal file
25
wiki/entities/F5TTS.md
Normal file
@@ -0,0 +1,25 @@
|
||||
---
|
||||
id: f5-tts
|
||||
title: "F5-TTS"
|
||||
type: entity
|
||||
tags: [AI, TTS, voice-cloning, open-source]
|
||||
sources:
|
||||
- "[[AI配音与声音克隆工具合集]]"
|
||||
last_updated: 2025-03-06
|
||||
---
|
||||
|
||||
## Overview
|
||||
|
||||
F5-TTS是开源语音克隆项目,2秒即可克隆声音,适合技术流用户,需要代码基础。
|
||||
|
||||
## Key Features
|
||||
|
||||
- 开源免费
|
||||
- 2秒克隆
|
||||
- 技术流首选
|
||||
- 需代码基础
|
||||
|
||||
## Connections
|
||||
- [[AI配音]] ← uses ← [[F5-TTS]]
|
||||
- [[声音克隆]] ← uses ← [[F5-TTS]]
|
||||
- [[开源工具]] ← extends ← [[F5-TTS]]
|
||||
24
wiki/entities/LangChain.md
Normal file
24
wiki/entities/LangChain.md
Normal file
@@ -0,0 +1,24 @@
|
||||
---
|
||||
id: langchain
|
||||
title: "LangChain"
|
||||
type: entity
|
||||
tags: [LLM, framework, RAG]
|
||||
sources:
|
||||
- "[[RAG从入门到精通系列1:基础RAG]]"
|
||||
last_updated: 2025-12-18
|
||||
---
|
||||
|
||||
## Overview
|
||||
|
||||
LangChain是RAG实现框架,用于构建基于LLM的应用。
|
||||
|
||||
## Key Features
|
||||
|
||||
- RAG流程封装
|
||||
- 多种数据源连接
|
||||
- Chain构建能力
|
||||
- Agent支持
|
||||
|
||||
## Connections
|
||||
- [[RAG]] ← uses ← [[LangChain]]
|
||||
- [[LLM应用开发]] ← uses ← [[LangChain]]
|
||||
45
wiki/entities/MCP.md
Normal file
45
wiki/entities/MCP.md
Normal file
@@ -0,0 +1,45 @@
|
||||
---
|
||||
title: "MCP"
|
||||
type: entity
|
||||
tags: [mcp, protocol, integration]
|
||||
last_updated: 2026-04-14
|
||||
---
|
||||
|
||||
# MCP
|
||||
|
||||
Modal Context Protocol(模态上下文协议),AI大模型与外围服务集成的协议。
|
||||
|
||||
## 基本信息
|
||||
- **全称**:Modal Context Protocol
|
||||
- **类型**:Client-Server架构协议
|
||||
- **用途**:实现大模型与外围工具服务的高效集成
|
||||
|
||||
## 核心功能
|
||||
|
||||
### 三种接口
|
||||
MCP Server提供三种功能接口:
|
||||
1. **资源获取**(Resource/GET):类似HTTP GET请求
|
||||
2. **工具调用**(Tool/POST):类似POST请求
|
||||
3. **Promise提示词**:用于多样化交互与扩展
|
||||
|
||||
### 客户端
|
||||
- [[Cursor]]:AI代码编辑器
|
||||
- Claude Desktop
|
||||
- 其他支持MCP的客户端
|
||||
|
||||
### 服务端
|
||||
- 热点新闻MCP Server
|
||||
- Sequential Thinking工具
|
||||
- [[n8n-mcp]]:N8N的MCP实现
|
||||
|
||||
## 接入方式
|
||||
1. **SSE服务方式**:通过Server-Sent Events接入
|
||||
2. **本地Command方式**:通过本地执行命令接入
|
||||
|
||||
## 使用场景
|
||||
|
||||
详见:
|
||||
- [[MCP在Cursor中的集成与应用]]
|
||||
|
||||
## 相关链接
|
||||
- [MCP协议文档](https://modelcontextprotocol.io/)
|
||||
45
wiki/entities/N8N.md
Normal file
45
wiki/entities/N8N.md
Normal file
@@ -0,0 +1,45 @@
|
||||
---
|
||||
title: "N8N"
|
||||
type: entity
|
||||
tags: [n8n, automation, workflow]
|
||||
last_updated: 2026-04-14
|
||||
---
|
||||
|
||||
# N8N
|
||||
|
||||
开源工作流自动化平台,支持节点连接执行任务。
|
||||
|
||||
## 基本信息
|
||||
- **类型**:工作流自动化平台
|
||||
- **开源协议**:Apache-2.0
|
||||
- **部署方式**:Docker、本地安装、云端
|
||||
- **特点**:可视化拖拽界面
|
||||
|
||||
## 核心概念
|
||||
|
||||
### 节点(Node)
|
||||
工作流中的单个操作单元,分为五类:
|
||||
- 触发器节点(Trigger)
|
||||
- 动作节点(Action)
|
||||
- 工具节点(Utility)
|
||||
- 代码节点(Code)
|
||||
- 高级AI节点(Advanced AI)
|
||||
|
||||
### 工作流(Workflow)
|
||||
由多个节点按一定顺序执行的自动化流程。
|
||||
|
||||
### 与AI Agent集成
|
||||
- [[n8n-mcp]]:N8N的MCP服务器实现
|
||||
- [[OpenClaw + n8n工作流编排]]:通过webhook安全集成
|
||||
- [[使用Claude自动生成N8N工作流]]
|
||||
|
||||
## 使用场景
|
||||
|
||||
详见各use case:
|
||||
- [[N8N Telegram Trigger配置]]
|
||||
- [[N8N全教程构建AI Agent]]
|
||||
- [[N8N工作流编排]]
|
||||
|
||||
## 相关链接
|
||||
- [N8N官网](https://n8n.io/)
|
||||
- [N8N文档](https://docs.n8n.io/)
|
||||
32
wiki/entities/NanoBanana.md
Normal file
32
wiki/entities/NanoBanana.md
Normal file
@@ -0,0 +1,32 @@
|
||||
---
|
||||
id: nano-banana
|
||||
title: "Nano Banana"
|
||||
type: entity
|
||||
tags: [google, AI, image-generation]
|
||||
sources:
|
||||
- "[[Nano Banana提示词框架]]"
|
||||
- "[[Nano Banana Pro提示词指南]]"
|
||||
- "[[Nano Banana 2使用指南]]"
|
||||
last_updated: 2025-12-01
|
||||
---
|
||||
|
||||
## Overview
|
||||
|
||||
Nano Banana是Google开发的AI图像生成模型,基于Gemini系列模型的图像生成能力。最新版本为Nano Banana 2(Gemini 3 Pro Image),是推理模型,生成前会进行内部推理。
|
||||
|
||||
## Key Features
|
||||
|
||||
- **推理模型**:生成前会进行内部推理,提升质量
|
||||
- **多分辨率支持**:1K、2K、4K分辨率
|
||||
- **多图像组合**:最多14张输入图像组合输出
|
||||
- **多语言长文本渲染**:擅长在中英文环境中渲染长文本
|
||||
|
||||
## Versions
|
||||
|
||||
- **Nano Banana**:基础版本
|
||||
- **Nano Banana Pro**:增强版本
|
||||
- **Nano Banana 2**:Gemini 3 Pro Image,推理模型
|
||||
|
||||
## Connections
|
||||
- [[AI图像生成]] ← uses ← [[Nano Banana]]
|
||||
- [[Nano Banana]] ← supports ← [[提示词框架]]
|
||||
36
wiki/entities/NotebookLM.md
Normal file
36
wiki/entities/NotebookLM.md
Normal file
@@ -0,0 +1,36 @@
|
||||
---
|
||||
id: notebooklm
|
||||
title: "NotebookLM"
|
||||
type: entity
|
||||
tags: [google, AI, learning, productivity]
|
||||
sources:
|
||||
- "[[7 ways I use NotebookLM to make my life easier]]"
|
||||
- "[[NotebookLM Open Source Alternatives]]"
|
||||
last_updated: 2025-11-23
|
||||
---
|
||||
|
||||
## Overview
|
||||
|
||||
NotebookLM是Google推出的AI学习和工作助手,核心特点是Source-Grounding机制确保回答准确且可溯源。
|
||||
|
||||
## Key Features
|
||||
|
||||
- **Source-Grounding**:限制知识库仅包含用户上传的文档,避免AI幻觉
|
||||
- **音频概览**:将文档转化为AI双人播客格式,适合被动学习
|
||||
- **项目中心**:集中管理研究、想法、会议记录的统一空间
|
||||
- **精确引用**:法律文档审查时提供精确引用
|
||||
|
||||
## Use Cases
|
||||
|
||||
- 信息整理和学习笔记
|
||||
- 文档问答和摘要
|
||||
- 项目研究和知识管理
|
||||
- 被动学习(音频概览)
|
||||
|
||||
## Aliases
|
||||
|
||||
- Google NotebookLM
|
||||
|
||||
## Connections
|
||||
- [[NotebookLM]] ← uses ← [[Source-Grounding]]
|
||||
- [[AI学习工具]] ← extends ← [[NotebookLM]]
|
||||
56
wiki/entities/OpenClaw.md
Normal file
56
wiki/entities/OpenClaw.md
Normal file
@@ -0,0 +1,56 @@
|
||||
---
|
||||
title: "OpenClaw"
|
||||
type: entity
|
||||
tags: [openclaw, agent, framework]
|
||||
last_updated: 2026-04-14
|
||||
---
|
||||
|
||||
# OpenClaw
|
||||
|
||||
开源AI Agent框架,支持多Agent协作、记忆系统、技能扩展。
|
||||
|
||||
## 基本信息
|
||||
- **类型**:AI Agent框架
|
||||
- **开源协议**:MIT
|
||||
- **平台**:跨平台(支持macOS、Linux、Windows)
|
||||
- **连接方式**:CLI、Telegram、Discord、WebUI
|
||||
|
||||
## 核心功能
|
||||
|
||||
### Workspace文件体系
|
||||
OpenClaw使用workspace目录下的文件来配置Agent行为:
|
||||
- [[AGENTS.md]]:Agent工作说明书
|
||||
- [[SOUL.md]]:Agent性格档案
|
||||
- [[USER.md]]:用户偏好
|
||||
- [[TOOLS.md]]:工具权限声明
|
||||
- [[IDENTITY.md]]:Agent身份元数据
|
||||
- [[BOOTSTRAP.md]]:首次启动引导
|
||||
- [[memory/]]:长期记忆目录
|
||||
|
||||
### 多Agent支持
|
||||
- 通过`sessions_spawn`启动子Agent
|
||||
- 通过`sessions_send`发送消息
|
||||
- 支持多Agent团队协作(如Milo、Josh等)
|
||||
|
||||
### 技能系统
|
||||
- 技能包目录:`skills/`
|
||||
- 支持第三方技能安装
|
||||
- MCP协议集成
|
||||
|
||||
### 记忆系统
|
||||
- 内置Markdown文件记忆
|
||||
- 支持向量语义搜索(通过memsearch)
|
||||
- 长期上下文保留
|
||||
|
||||
## 使用场景
|
||||
|
||||
详见各use case:
|
||||
- [[多Agent专业团队]]
|
||||
- [[自愈家庭服务器]]
|
||||
- [[第二大脑]]
|
||||
- [[目标驱动自主任务]]
|
||||
|
||||
## 相关链接
|
||||
- [OpenClaw GitHub](https://github.com/openclaw/openclaw)
|
||||
- [OpenClaw文档](https://docs.openclaw.ai)
|
||||
- [OpenClaw Showcase](https://openclaw.ai/showcase)
|
||||
24
wiki/entities/剪映.md
Normal file
24
wiki/entities/剪映.md
Normal file
@@ -0,0 +1,24 @@
|
||||
---
|
||||
id: jianying
|
||||
title: "剪映"
|
||||
type: entity
|
||||
tags: [video-editing, ByteDance, TikTok]
|
||||
sources:
|
||||
- "[[AI配音与声音克隆工具合集]]"
|
||||
last_updated: 2025-03-06
|
||||
---
|
||||
|
||||
## Overview
|
||||
|
||||
剪映是字节跳动推出的视频编辑工具,抖音官方,短视频首选,部分音色需VIP。
|
||||
|
||||
## Key Features
|
||||
|
||||
- 抖音官方工具
|
||||
- 短视频首选
|
||||
- 部分音色需VIP
|
||||
- 内置AI配音功能
|
||||
|
||||
## Connections
|
||||
- [[AI视频制作]] ← uses ← [[剪映]]
|
||||
- [[短视频制作]] ← uses ← [[剪映]]
|
||||
24
wiki/entities/海螺AI.md
Normal file
24
wiki/entities/海螺AI.md
Normal file
@@ -0,0 +1,24 @@
|
||||
---
|
||||
id: hailo-ai
|
||||
title: "海螺AI"
|
||||
type: entity
|
||||
tags: [AI, TTS, voice-cloning, MiniMax]
|
||||
sources:
|
||||
- "[[AI配音与声音克隆工具合集]]"
|
||||
last_updated: 2025-03-06
|
||||
---
|
||||
|
||||
## Overview
|
||||
|
||||
海螺AI是MiniMax出品的AI工具,主打语音克隆和配音功能,对中文支持好,小白友好。
|
||||
|
||||
## Key Features
|
||||
|
||||
- 30秒克隆声音
|
||||
- 免费使用
|
||||
- 中文支持好
|
||||
- 小白友好界面
|
||||
|
||||
## Connections
|
||||
- [[AI配音]] ← uses ← [[海螺AI]]
|
||||
- [[声音克隆]] ← uses ← [[海螺AI]]
|
||||
123
wiki/index.md
Normal file
123
wiki/index.md
Normal file
@@ -0,0 +1,123 @@
|
||||
# Wiki Index
|
||||
|
||||
## Overview
|
||||
- [Overview](overview.md)
|
||||
|
||||
## Sources
|
||||
- [万字讲透OpenClaw-Workspace深度解析](sources/wan-zi-jiang-tou-openclaw-workspace-shen-du-jie-xi.md)
|
||||
- [万字保姆级教程-90天跑通一人公司模式](sources/90-tian-pao-tong-yi-ren-gong-si-mo-shi.md)
|
||||
- [MCP在Cursor中的集成与应用详解](sources/mcp-zai-cursor-zhong-de-ji-cheng-yu-ying-yong.md)
|
||||
- [N8N Full Tutorial Building AI Agents](sources/n8n-full-tutorial-building-ai-agents.md)
|
||||
- [N8N Configure Telegram Trigger](sources/n8n-configure-telegram-trigger.md)
|
||||
- [使用Claude自动生成N8N工作流的实操教程](sources/shi-yong-claude-zi-dong-sheng-cheng-n8n-gong-zuo-liu.md)
|
||||
- [N8N+Claude通过自然语言自动化工作流](sources/n8n-claude-zi-dong-hua-gong-zuo-liu.md)
|
||||
- [Podcast Production Pipeline](sources/podcast-production-pipeline.md)
|
||||
- [Automated Meeting Notes & Action Items](sources/meeting-notes-action-items.md)
|
||||
- [Daily YouTube Digest](sources/daily-youtube-digest.md)
|
||||
- [Multi-Agent Content Factory](sources/content-factory.md)
|
||||
- [Self-Healing Home Server](sources/self-healing-home-server.md)
|
||||
- [Health & Symptom Tracker](sources/health-symptom-tracker.md)
|
||||
- [Project State Management System](sources/project-state-management.md)
|
||||
- [Multi-Agent Specialized Team](sources/multi-agent-team.md)
|
||||
- [AI-Powered Earnings Tracker](sources/earnings-tracker.md)
|
||||
- [Multi-Channel Personal Assistant](sources/multi-channel-assistant.md)
|
||||
- [Event Guest Confirmation](sources/event-guest-confirmation.md)
|
||||
- [Market Research & Product Factory](sources/market-research-product-factory.md)
|
||||
- [Custom Morning Brief](sources/custom-morning-brief.md)
|
||||
- [Inbox De-clutter](sources/inbox-declutter.md)
|
||||
- [Daily Reddit Digest](sources/daily-reddit-digest.md)
|
||||
- [Autonomous Project Management with Subagents](sources/autonomous-project-management.md)
|
||||
- [Pre-Build Idea Validator](sources/pre-build-idea-validator.md)
|
||||
- [Dynamic Dashboard with Sub-agent Spawning](sources/dynamic-dashboard.md)
|
||||
- [Todoist Task Manager](sources/todoist-task-manager.md)
|
||||
- [Habit Tracker & Accountability Coach](sources/habit-tracker-accountability-coach.md)
|
||||
- [LaTeX Paper Writing](sources/latex-paper-writing.md)
|
||||
- [Second Brain](sources/second-brain.md)
|
||||
- [Multi-Channel Customer Service](sources/multi-channel-customer-service.md)
|
||||
- [OpenClaw + n8n Workflow Orchestration](sources/n8n-workflow-orchestration.md)
|
||||
- [Local CRM Framework with DenchClaw](sources/local-crm-framework.md)
|
||||
- [Overnight Mini App Builder](sources/overnight-mini-app-builder.md)
|
||||
- [Polymarket Autopilot](sources/polymarket-autopilot.md)
|
||||
- [YouTube Content Pipeline](sources/youtube-content-pipeline.md)
|
||||
- [Personal CRM with Automatic Contact Discovery](sources/personal-crm.md)
|
||||
- [Knowledge Base RAG](sources/knowledge-base-rag.md)
|
||||
- [X/Twitter Automation from Chat](sources/x-twitter-automation.md)
|
||||
- [Multi-Source Tech News Digest](sources/multi-source-tech-news-digest.md)
|
||||
- [Family Calendar Aggregation & Household Assistant](sources/family-calendar-household-assistant.md)
|
||||
- [OpenClaw as Desktop Cowork (AionUi)](sources/aionui-cowork-desktop.md)
|
||||
- [Semantic Memory Search](sources/semantic-memory-search.md)
|
||||
- [arXiv Paper Reader](sources/arxiv-paper-reader.md)
|
||||
- [Autonomous Educational Game Development Pipeline](sources/autonomous-game-dev-pipeline.md)
|
||||
- [Phone Call Notifications](sources/phone-call-notifications.md)
|
||||
- [X Account Analysis](sources/x-account-analysis.md)
|
||||
- [7 ways I use NotebookLM to make my life easier](sources/7-ways-notebooklm.md)
|
||||
- [Designing for Agentic AI](sources/designing-for-agentic-ai.md)
|
||||
- [Multi-Agent System Reliability](sources/multi-agent-system-reliability.md)
|
||||
- [A Formalization of Recursive Self-Optimizing Generative Systems](sources/a-formalization-of-recursive-self-optimizing-generative-systems.md)
|
||||
- [14个免费AI图生视频工具](sources/14-ge-mian-fei-de-ai-tu-sheng-shi-ping-gong-ju.md)
|
||||
- [2025年11个神级AI开源平替](sources/2025-nian-11-ge-shen-ji-ai-kai-yuan-ping-ti.md)
|
||||
- [Claude Skills最值得研究的AI范式](sources/claude-skills.md)
|
||||
- [Best 7 News APIs](sources/best-7-news-api.md)
|
||||
- [Vibe Coding Guide](sources/vibe-coding-guide.md)
|
||||
- [NotebookLM Open Source Alternatives](sources/notebooklm-open-source-alternatives.md)
|
||||
- [YouTube RSS Feed](sources/youtube-rss-feed.md)
|
||||
- [LLMs RAG AI Agent区别](sources/llms-rag-ai-agent-qu-bie.md)
|
||||
- [Nano Banana提示词框架](sources/nano-banana-prompting-guide.md)
|
||||
- [Nano Banana Pro提示词指南](sources/nano-banana-pro-prompting-guide.md)
|
||||
- [Multiple Interests Guide](sources/multiple-interests-guide.md)
|
||||
- [Never Write Another Prompt](sources/never-write-another-prompt.md)
|
||||
- [The Picture They Paint of You](sources/the-picture-they-paint-of-you.md)
|
||||
- [RAG从入门到精通系列1:基础RAG](sources/rag-cong-ru-men-dao-jing-tong.md)
|
||||
- [OpenAI ChatGPT个性化定义](sources/openai-chatgpt-ge-xing-hua-ding-yi.md)
|
||||
- [Build Your Own X](sources/build-your-own-x.md)
|
||||
- [抑郁典型梦中人](sources/yi-yu-dian-xing-meng-zhong-ren.md)
|
||||
- [Gemini Product Manager PRD](sources/gemini-product-manager-prd.md)
|
||||
- [AI配音与声音克隆工具合集](sources/ai-pei-yin-sheng-yin-kelong-gong-ju.md)
|
||||
- [固定镜头短视频AI全流程制作](sources/gu-ding-jing-tou-duan-shi-pin-ai-sheng-chan.md)
|
||||
- [大模型相关术语框架总结](sources/llm-terms-framework.md)
|
||||
- [Nano Banana 2使用指南](sources/nano-banana-2-guide.md)
|
||||
- [如何写出完美的Prompt](sources/perfect-prompt-guide.md)
|
||||
|
||||
## Entities
|
||||
- [OpenClaw](entities/OpenClaw.md)
|
||||
- [N8N](entities/N8N.md)
|
||||
- [MCP](entities/MCP.md)
|
||||
- [NotebookLM](entities/NotebookLM.md)
|
||||
- [NanoBanana](entities/NanoBanana.md)
|
||||
- [DeepSider](entities/DeepSider.md)
|
||||
- [ElevenLabs](entities/ElevenLabs.md)
|
||||
- [海螺AI](entities/海螺AI.md)
|
||||
- [F5-TTS](entities/F5TTS.md)
|
||||
- [剪映](entities/剪映.md)
|
||||
- [LangChain](entities/LangChain.md)
|
||||
|
||||
## Concepts
|
||||
- [Workspace](concepts/Workspace.md)
|
||||
- [多Agent系统](concepts/多Agent系统.md)
|
||||
- [工作流自动化](concepts/工作流自动化.md)
|
||||
- [记忆系统](concepts/记忆系统.md)
|
||||
- [技能系统](concepts/技能系统.md)
|
||||
- [Source-Grounding](concepts/SourceGrounding.md)
|
||||
- [音频概览](concepts/音频概览.md)
|
||||
- [提示词框架](concepts/提示词框架.md)
|
||||
- [GenAI](concepts/GenAI.md)
|
||||
- [Agentic AI](concepts/AgenticAI.md)
|
||||
- [RAG](concepts/RAG.md)
|
||||
- [Embedding](concepts/Embedding.md)
|
||||
- [Prompt能力](concepts/Prompt能力.md)
|
||||
- [需求拆解](concepts/需求拆解.md)
|
||||
- [结构化表达](concepts/结构化表达.md)
|
||||
- [思维链引导](concepts/思维链引导.md)
|
||||
- [AI配音](concepts/AI配音.md)
|
||||
- [声音克隆](concepts/声音克隆.md)
|
||||
- [固定机位](concepts/固定机位.md)
|
||||
- [首尾针动画](concepts/首尾针动画.md)
|
||||
- [九宫格法](concepts/九宫格法.md)
|
||||
- [LLM](concepts/LLM.md)
|
||||
- [Token](concepts/Token.md)
|
||||
- [MCP](concepts/MCP.md)
|
||||
- [Agent](concepts/Agent.md)
|
||||
- [vLLM](concepts/vLLM.md)
|
||||
- [Claude Skills](concepts/ClaudeSkills.md)
|
||||
- [流程工程](concepts/流程工程.md)
|
||||
- [Vibe Coding](concepts/VibeCoding.md)
|
||||
40
wiki/log.md
Normal file
40
wiki/log.md
Normal file
@@ -0,0 +1,40 @@
|
||||
# Wiki Log
|
||||
|
||||
## [2026-04-14] ingest | AI目录批量摄取(26个文件)
|
||||
|
||||
### 摄取统计
|
||||
- Source页面:26个
|
||||
- Entity页面:11个(NotebookLM、NanoBanana、DeepSider、ElevenLabs、海螺AI、F5-TTS、剪映、LangChain等)
|
||||
- Concept页面:21个(Source-Grounding、音频概览、提示词框架、GenAI、Agentic AI、RAG、Embedding、Prompt能力、AI配音、声音克隆、固定机位、首尾针动画、九宫格法、LLM、Token、MCP、Agent、vLLM、Claude Skills、流程工程、Vibe Coding等)
|
||||
|
||||
### 主要主题
|
||||
- AI图像生成:Nano Banana提示词工程
|
||||
- AI配音与声音克隆:ElevenLabs、海螺AI、F5-TTS
|
||||
- 视频制作:固定镜头、九宫格法、首尾针动画
|
||||
- RAG检索增强生成
|
||||
- 大模型术语体系:LLM、MCP、Agent、vLLM、Token
|
||||
- Claude Skills与流程工程
|
||||
- Prompt能力与提示词工程
|
||||
|
||||
### 冲突检测
|
||||
- 无重大冲突发现
|
||||
|
||||
## [2026-04-14] ingest | Agent目录批量摄取(45个文件)
|
||||
|
||||
### 摄取统计
|
||||
- Source页面:45个
|
||||
- Entity页面:3个(OpenClaw、N8N、MCP)
|
||||
- Concept页面:5个(Workspace、多Agent系统、工作流自动化、记忆系统、技能系统)
|
||||
|
||||
### 来源文件
|
||||
- 主目录:8个文件
|
||||
- usecases子目录:37个文件
|
||||
|
||||
### 主要主题
|
||||
- OpenClaw配置与使用
|
||||
- N8N工作流自动化
|
||||
- MCP协议集成
|
||||
- 多Agent协作系统
|
||||
- 个人效率工具
|
||||
- 内容创作自动化
|
||||
- 商业应用(CRM、市场研究)
|
||||
114
wiki/overview.md
Normal file
114
wiki/overview.md
Normal file
@@ -0,0 +1,114 @@
|
||||
# Overview
|
||||
|
||||
本Wiki涵盖两大核心领域:**Agent(智能体)系统**和**AI应用**。Agent系统以OpenClaw、N8N、MCP为核心;AI应用涵盖图像生成、配音克隆、视频制作、RAG检索等场景。
|
||||
|
||||
## 核心主题
|
||||
|
||||
### Agent系统
|
||||
- **OpenClaw**:开源AI Agent框架,支持多Agent协作、记忆系统、技能扩展
|
||||
- **N8N**:开源工作流自动化平台,可与AI Agent结合实现复杂自动化
|
||||
- **MCP (Modal Context Protocol)**:AI大模型与外围服务集成的协议
|
||||
- **多Agent系统**:多个专业Agent协同工作的架构模式
|
||||
|
||||
### AI应用
|
||||
- **图像生成**:Nano Banana、Midjourney等模型的提示词工程
|
||||
- **AI配音与声音克隆**:ElevenLabs、海螺AI、F5-TTS等工具
|
||||
- **视频制作**:固定镜头短视频AI全流程、九宫格法、首尾针动画
|
||||
- **RAG检索**:检索增强生成解决LLM幻觉问题
|
||||
|
||||
## 问题域
|
||||
|
||||
### Agent系统
|
||||
- Agent开发与配置(Workspace文件体系)
|
||||
- 工作流自动化(N8N节点和工作流)
|
||||
- 跨平台集成(Telegram、Discord、Slack等)
|
||||
- 个人效率提升(习惯追踪、日历管理、邮件处理)
|
||||
- 内容创作自动化(YouTube、Twitter、播客等)
|
||||
- 商业应用(CRM、市场研究、财报追踪)
|
||||
|
||||
### AI应用
|
||||
- 提示词工程:角色-需求-场景-目标四要素结构
|
||||
- 图像生成质量控制:物件描述框架、人物描述框架
|
||||
- 声音克隆与配音:多语言TTS、各平台特点
|
||||
- 视频内容创作:分镜拆解、一致性保证
|
||||
- 大模型知识体系:LLM、MCP、Agent、RAG、vLLM、Token
|
||||
|
||||
## 方法/机制
|
||||
|
||||
### Agent系统
|
||||
- **Workspace文件体系**(AGENTS.md、SOUL.md、USER.md等):定义Agent行为
|
||||
- **Sub-agent多Agent协作**:分散式协调通过共享状态文件
|
||||
- **Cron定时任务与心跳机制**:自动化定期执行
|
||||
- **MCP协议工具调用**:Client-Server架构的服务集成
|
||||
- **Webhook安全集成**:OpenClaw与N8N的安全集成模式
|
||||
- **向量语义搜索**:memsearch实现记忆的语义检索
|
||||
|
||||
### AI应用
|
||||
- **提示词框架**:物件描述、人物描述的JSON标准化结构
|
||||
- **Source-Grounding**:NotebookLM限制知识库确保回答准确性
|
||||
- **九宫格法**:一次性生成多个分镜保证画面一致性
|
||||
- **首尾针动画**:AI补齐中间帧实现平滑过渡
|
||||
- **RAG三步流程**:索引→检索→生成
|
||||
|
||||
## 关键实体
|
||||
|
||||
### Agent系统
|
||||
- [[OpenClaw]]:开源AI Agent框架
|
||||
- [[N8N]]:工作流自动化平台
|
||||
- [[MCP]]:模态上下文协议
|
||||
- [[DenchClaw]]:本地CRM框架
|
||||
- [[AionUi]]:桌面Cowork应用
|
||||
|
||||
### AI应用
|
||||
- [[NotebookLM]]:Google AI学习工具
|
||||
- [[Nano Banana]]:Google AI图像生成模型
|
||||
- [[DeepSider]]:多AI模型聚合浏览器插件
|
||||
- [[ElevenLabs]]:国际AI配音平台
|
||||
- [[海螺AI]]:MiniMax出品的AI工具
|
||||
- [[F5-TTS]]:开源语音克隆项目
|
||||
- [[剪映]]:字节跳动视频编辑工具
|
||||
|
||||
## 关键概念
|
||||
|
||||
### Agent系统
|
||||
- [[Workspace]]:Agent的工作台目录配置体系
|
||||
- [[多Agent系统]]:多个专业Agent协同工作的架构
|
||||
- [[工作流自动化]]:使用工具自动执行重复性任务
|
||||
- [[记忆系统]]:Agent跨会话保留上下文的能力
|
||||
- [[技能系统]]:OpenClaw的扩展机制
|
||||
- [[n8n-mcp]]:N8N的MCP服务器实现
|
||||
- [[idea-reality-mcp]]:预构建创意验证MCP
|
||||
|
||||
### AI应用
|
||||
- [[Prompt能力]]:清晰界定需求+结构化思维表达
|
||||
- [[需求拆解]]:将模糊目标转化为具体可执行子任务
|
||||
- [[结构化表达]]:用清晰逻辑组织信息
|
||||
- [[思维链引导]]:让AI逐步推理
|
||||
- [[Source-Grounding]]:限制知识库确保AI回答准确性
|
||||
- [[音频概览]]:将文档转化为AI双人播客格式
|
||||
- [[提示词框架]]:结构化描述图像生成需求的模板
|
||||
- [[AI配音]]:文本转语音技术
|
||||
- [[声音克隆]]:用少量样本重建个人声音
|
||||
- [[固定机位]]:摄像机位置固定不变的拍摄方式
|
||||
- [[首尾针动画]]:通过首尾帧AI自动补齐中间动作
|
||||
- [[九宫格法]]:一次性生成3x3共九个分镜画面
|
||||
- [[RAG]]:检索增强生成
|
||||
- [[Embedding]]:将文本转换为数值向量的技术
|
||||
- [[LLM]]:大语言模型
|
||||
- [[Agent]]:智能体
|
||||
- [[vLLM]]:高效LLM推理框架
|
||||
- [[Token]]:大模型基本输入单元
|
||||
|
||||
## Source分类
|
||||
|
||||
### Agent系统(45个)
|
||||
涵盖:OpenClaw Workspace、N8N工作流、MCP协议、多Agent协作、记忆系统、技能扩展等
|
||||
|
||||
### AI应用(26个)
|
||||
涵盖:NotebookLM、Claude Skills、Nano Banana提示词、RAG基础、AI配音工具、视频制作流程、大模型术语等
|
||||
|
||||
## 来源分布
|
||||
- 微信公众号:Agent系统相关内容
|
||||
- YouTube视频:部分AI工具教程
|
||||
- OpenClaw Use Cases:37个
|
||||
- 技术文档:AI图像生成、RAG、提示词工程等
|
||||
@@ -0,0 +1,47 @@
|
||||
---
|
||||
title: "14个免费的AI图生视频工具"
|
||||
type: source
|
||||
tags: [AI, image-to-video, tools]
|
||||
date: 2025-12-05
|
||||
---
|
||||
|
||||
## Source File
|
||||
- [[raw/AI/14个免费的AI图生视频工具,用AI让图片动起来 - AI视频教程 AI自动化工作流定制服务 AI培训学习平台 黑喵大叔.md]]
|
||||
|
||||
## Summary
|
||||
- 核心主题:14个免费AI图生视频工具评测
|
||||
- 问题域:视频制作需要专业设备和技能,门槛高
|
||||
- 方法/机制:上传静态图片,AI分析内容并生成动态视频
|
||||
- 结论/价值:AI图生视频工具降低视频创作门槛,实现便捷创作
|
||||
|
||||
## Key Claims
|
||||
- 绘蛙AI视频(阿里):模特图转视频,支持多格式高分辨率
|
||||
- 智谱清影:30秒生成6秒高清视频,支持音效匹配
|
||||
- 通义万相(阿里):支持提示词控制运动,匹配音效
|
||||
- Vidu(清华):全球首个"多主体参考"功能,10秒生成视频
|
||||
- 可灵AI(快手):1080p分辨率,真实物理规律表现
|
||||
- 海螺AI(MiniMax):主体一致性优秀,支持多种艺术风格
|
||||
|
||||
## Key Concepts
|
||||
- [[图生视频]]:将静态图片转化为动态视频的AI技术
|
||||
- [[主体一致性]]:视频中角色/物体保持一致的能力
|
||||
- [[运动控制]]:通过文本提示词控制视频中主体的运动方式
|
||||
|
||||
## Key Entities
|
||||
- [[绘蛙AI视频]]:阿里巴巴AI图生视频工具
|
||||
- [[智谱清影]]:智谱AI视频生成工具
|
||||
- [[通义万相]]:阿里巴巴AI视频生成工具
|
||||
- [[Vidu]]:生数科技与清华大学发布的视频大模型
|
||||
- [[可灵AI]]:快手AI创作平台
|
||||
- [[海螺AI]]:MiniMax推出的AI视频工具
|
||||
- [[即梦AI]]:字节跳动一站式AI创意创作平台
|
||||
- [[PixVerse]]:爱诗科技AI视频工具
|
||||
- [[Stable Video]]:Stability AI视频生成平台
|
||||
|
||||
## Connections
|
||||
- [[AI视频生成]] ← includes ← [[图生视频]]
|
||||
- [[阿里巴巴]] ← provides ← [[绘蛙AI视频]]
|
||||
- [[阿里巴巴]] ← provides ← [[通义万相]]
|
||||
- [[快手]] ← provides ← [[可灵AI]]
|
||||
|
||||
## Contradictions
|
||||
48
wiki/sources/2025-nian-11-ge-shen-ji-ai-kai-yuan-ping-ti.md
Normal file
48
wiki/sources/2025-nian-11-ge-shen-ji-ai-kai-yuan-ping-ti.md
Normal file
@@ -0,0 +1,48 @@
|
||||
---
|
||||
title: "2025年11个神级AI开源平替"
|
||||
type: source
|
||||
tags: [open-source, AI, GitHub]
|
||||
date: 2026-01-01
|
||||
---
|
||||
|
||||
## Source File
|
||||
- [[raw/AI/2025 年 11 个神级 AI 开源平替,GitHub 杀疯了。.md]]
|
||||
|
||||
## Summary
|
||||
- 核心主题:2025年GitHub热门AI开源项目盘点
|
||||
- 问题域:闭源AI产品价格昂贵,开源替代品需求旺盛
|
||||
- 方法/机制:按类别梳理各领域最火热的开源项目
|
||||
- 结论/价值:国产AI模型(DeepSeek、Qwen)在开源界表现亮眼
|
||||
|
||||
## Key Claims
|
||||
- 大语言模型:DeepSeek R1、Qwen 3为开源界标杆
|
||||
- AI生图:Flux(开源Midjourney)、Stable Diffusion(LoRA/ControlNet生态最丰富)
|
||||
- AI生视频:HunyuanVideo(腾讯)参数量最大,中文理解最强
|
||||
- 通用智能体:OpenManus(5万Star)为核心开源平替
|
||||
- AI Coding:Cline为Cursor最佳开源平替
|
||||
- 智能体工作流:n8n(16万Star)、Dify为最强开源项目
|
||||
|
||||
## Key Concepts
|
||||
- [[开源平替]]:开源替代闭源产品的方案
|
||||
- [[AI生图]]:开源模型Flux、Stable Diffusion
|
||||
- [[AI生视频]]:HunyuanVideo、Veo 3
|
||||
- [[AI智能体]]:Manus、OpenManus
|
||||
- [[AI编程]]:Cline、Claude Code
|
||||
|
||||
## Key Entities
|
||||
- [[DeepSeek]]:国产开源大模型
|
||||
- [[Qwen]]:通义千问开源模型
|
||||
- [[Flux]]:前SD团队开发的AI生图模型
|
||||
- [[HunyuanVideo]]:腾讯混元视频生成模型
|
||||
- [[OpenManus]]:通用智能体开源项目
|
||||
- [[Cline]]:VS Code AI编程插件
|
||||
- [[n8n]]:工作流自动化开源平台
|
||||
- [[Dify]]:LLM应用开发平台
|
||||
|
||||
## Connections
|
||||
- [[开源AI]] ← includes ← [[大语言模型]]
|
||||
- [[开源AI]] ← includes ← [[AI生图]]
|
||||
- [[开源AI]] ← includes ← [[AI生视频]]
|
||||
- [[GitHub]] ← hosts ← [[开源AI项目]]
|
||||
|
||||
## Contradictions
|
||||
35
wiki/sources/7-ways-notebooklm.md
Normal file
35
wiki/sources/7-ways-notebooklm.md
Normal file
@@ -0,0 +1,35 @@
|
||||
---
|
||||
title: "7 ways I use NotebookLM to make my life easier"
|
||||
type: source
|
||||
tags: [notebooklm, google, learning, productivity]
|
||||
date: 2025-11-23
|
||||
---
|
||||
|
||||
## Source File
|
||||
- [[raw/AI/7 ways I use NotebookLM to make my life easier.md]]
|
||||
|
||||
## Summary
|
||||
- 核心主题:NotebookLM使用技巧
|
||||
- 问题域:信息过载、文档处理、学习效率
|
||||
- 方法/机制:source-grounding文档问答、音频概览、项目管理
|
||||
- 结论/价值:NotebookLM通过严格限制知识库范围保证准确性,是个人学习和项目管理的强大助手
|
||||
|
||||
## Key Claims
|
||||
- NotebookLM的source-grounding机制确保回答准确且可溯源
|
||||
- 音频概览功能将文档转化为双人播客,适合被动学习
|
||||
- 可作为个性化项目管理中心,集中管理分散的研究和想法
|
||||
- 法律文档审查时提供精确引用,避免AI幻觉
|
||||
|
||||
## Key Concepts
|
||||
- [[Source-Grounding]]:NotebookLM的核心机制,限制知识库仅包含用户上传的文档
|
||||
- [[音频概览]]:将文档转化为AI双人播客格式
|
||||
- [[项目中心]]:集中管理研究、想法、会议记录的统一空间
|
||||
|
||||
## Key Entities
|
||||
- [[NotebookLM]]:Google推出的AI学习工具
|
||||
|
||||
## Connections
|
||||
- [[NotebookLM]] ← uses ← [[Source-Grounding]]
|
||||
- [[AI学习工具]] ← extends ← [[NotebookLM]]
|
||||
|
||||
## Contradictions
|
||||
36
wiki/sources/90-tian-pao-tong-yi-ren-gong-si-mo-shi.md
Normal file
36
wiki/sources/90-tian-pao-tong-yi-ren-gong-si-mo-shi.md
Normal file
@@ -0,0 +1,36 @@
|
||||
---
|
||||
title: "万字保姆级教程-90天跑通一人公司模式"
|
||||
type: source
|
||||
tags: [一人公司, 个人品牌, ikigai, 商业变现]
|
||||
date: 2026-03-29
|
||||
---
|
||||
|
||||
## Source File
|
||||
- [[raw/Agent/万字保姆级教程-90天跑通一人公司模式-2026-03-29.md]]
|
||||
|
||||
## Summary
|
||||
- 核心主题:用AI提示词和工具在90天内跑通一人公司模式
|
||||
- 问题域:个人如何找到自己的优势并实现商业变现
|
||||
- 方法/机制:Ikigai模型(热爱、擅长、市场需求、报酬四圈交集)、AI辅助内容生产、产品体系搭建
|
||||
- 结论/价值:一人公司的关键在于更聪明地定位,而非更努力工作
|
||||
|
||||
## Key Claims
|
||||
- 天才地带理论:找到能产生心流的活动区域
|
||||
- 底层能力三问:追溯童年、毫不费力、底层通用
|
||||
- 四个心理陷阱:愧疚陷阱、效率陷阱、卓越陷阱、努力陷阱
|
||||
- Ikigai是热情、天职、使命、职业的交汇点
|
||||
- 产品体系四层:引流产品、入门产品、核心产品、高价产品
|
||||
|
||||
## Key Concepts
|
||||
- [[Ikigai]]:日本概念,指生命的意义,四个圆圈的交集
|
||||
- [[一人公司]]:用最小杠杆撬动最大价值的商业模式
|
||||
- [[底层能力]]:藏在冰山下的通用能力
|
||||
- [[产品体系]]:从引流到高价的完整产品矩阵
|
||||
|
||||
## Key Entities
|
||||
- 盖伊·亨德里克斯:心理学家,提出天才地带概念
|
||||
- 张飞宇:营销人,公众号作者
|
||||
|
||||
## Connections
|
||||
- [[一人公司]] ← uses ← [[Ikigai]]
|
||||
- [[一人公司]] ← uses ← [[产品体系]]
|
||||
@@ -0,0 +1,36 @@
|
||||
---
|
||||
title: "A Formalization of Recursive Self-Optimizing Generative Systems"
|
||||
type: source
|
||||
tags: [recursive-optimization, formal-model, meta-learning, prompt-engineering]
|
||||
date: 2025-12-30
|
||||
---
|
||||
|
||||
## Source File
|
||||
- [[raw/AI/A Formalization of Recursive Self-Optimizing Generative Systems.md]]
|
||||
|
||||
## Summary
|
||||
- 核心主题:递归自优化生成系统的形式化模型
|
||||
- 问题域:如何形式化描述AI系统通过迭代自我修改构建稳定生成能力
|
||||
- 方法/机制:自映射、固定点结构、λ演算表述
|
||||
- 结论/价值:递归自优化自然引导到固定点结构而非终端输出
|
||||
|
||||
## Key Claims
|
||||
- 系统目标不是直接产生最优输出,而是通过迭代自修改构建稳定生成能力
|
||||
- 稳定生成能力定义为生成器空间上自映射的固定点
|
||||
- 递归结构可使用无类型λ演算表达
|
||||
- bootstrapping元生成过程由固定点语义控制
|
||||
|
||||
## Key Concepts
|
||||
- [[递归自优化]]:通过迭代生成-优化-更新循环自我完善
|
||||
- [[固定点]]:生成器在自映射下的不变状态
|
||||
- [[自映射]]:生成器空间到自身的映射
|
||||
- [[元生成]]:生成器更新生成器本身的过程
|
||||
- [[Bootstrap]]:从初始版本启动递归优化的起点
|
||||
|
||||
## Key Entities
|
||||
|
||||
## Connections
|
||||
- [[自优化AI]] ← formalizes ← [[递归自优化系统]]
|
||||
- [[固定点语义]] ← governs ← [[Bootstrap]]
|
||||
|
||||
## Contradictions
|
||||
41
wiki/sources/ai-pei-yin-sheng-yin-kelong-gong-ju.md
Normal file
41
wiki/sources/ai-pei-yin-sheng-yin-kelong-gong-ju.md
Normal file
@@ -0,0 +1,41 @@
|
||||
---
|
||||
title: "AI配音与声音克隆工具合集"
|
||||
type: source
|
||||
tags: [AI配音, 声音克隆, TTS]
|
||||
date: 2025-03-06
|
||||
---
|
||||
|
||||
## Source File
|
||||
- [[raw/AI/二创视频必不可少!2025年最热门AI工具推荐合集-AI配音、声音克隆.md]]
|
||||
|
||||
## Summary
|
||||
- 核心主题:AI配音和声音克隆工具评测
|
||||
- 问题域:二创视频需要高效配音解决方案
|
||||
- 方法/机制:评测主流AI配音工具的功能和价格
|
||||
- 结论/价值:不同场景有不同最优选择
|
||||
|
||||
## Key Claims
|
||||
- ElevenLabs:国际顶流,30+语言,支持情感变化,付费较贵
|
||||
- 海螺AI(MiniMax):小白友好,30秒克隆,免费,中文支持好
|
||||
- F5-TTS:开源免费,2秒克隆,技术流首选,需代码基础
|
||||
- TTSMaker:每周免费3万字,50+语言,300+音色
|
||||
- 剪映:抖音官方,短视频首选,部分音色需VIP
|
||||
- 魔音工坊:500+音色,企业首选
|
||||
- AnyVoice:3秒克隆,中英日韩,免费无限
|
||||
|
||||
## Key Concepts
|
||||
- [[AI配音]]:文本转语音技术
|
||||
- [[声音克隆]]:用少量样本重建个人声音
|
||||
- [[TTS]]:Text-to-Speech文字转语音
|
||||
|
||||
## Key Entities
|
||||
- [[ElevenLabs]]:国际AI配音平台
|
||||
- [[海螺AI]]:MiniMax出品的AI工具
|
||||
- [[F5-TTS]]:开源语音克隆项目
|
||||
- [[剪映]]:字节跳动视频编辑工具
|
||||
|
||||
## Connections
|
||||
- [[二创视频]] ← uses ← [[AI配音]]
|
||||
- [[内容创作]] ← uses ← [[声音克隆]]
|
||||
|
||||
## Contradictions
|
||||
34
wiki/sources/aionui-cowork-desktop.md
Normal file
34
wiki/sources/aionui-cowork-desktop.md
Normal file
@@ -0,0 +1,34 @@
|
||||
---
|
||||
title: "OpenClaw as Desktop Cowork (AionUi)"
|
||||
type: source
|
||||
tags: [openclaw, aionui, desktop, remote]
|
||||
date: 2026-03-06
|
||||
---
|
||||
|
||||
## Source File
|
||||
- [[raw/Agent/usecases/aionui-cowork-desktop.md]]
|
||||
|
||||
## Summary
|
||||
- 核心主题:使用AionUi桌面Cowork UI运行OpenClaw
|
||||
- 问题域:希望看到Agent正在做什么而非仅从日志推断
|
||||
- 方法/机制:AionUi提供桌面Cowork空间,远程通过Telegram或WebUI访问
|
||||
- 结论/价值:在桌面UI中使用OpenClaw,同时获得远程救援能力
|
||||
|
||||
## Key Claims
|
||||
- 真实桌面UI:看到OpenClaw读写文件、运行命令、浏览网页
|
||||
- 远程OpenClaw救援:当OpenClaw损坏或无法连接时,使用内置OpenClaw部署专家
|
||||
- 多Agent应用:一个应用中运行OpenClaw、内置Agent、Claude Code、Codex等
|
||||
- MCP一次配置,所有Agent同步:MCP服务器在AionUi中配置一次,sync到OpenClaw和其他Agent
|
||||
|
||||
## Key Insights
|
||||
- 当OpenClaw无法连接且你不在机器旁时:打开AionUi通过Telegram或WebUI,使用内置OpenClaw部署专家
|
||||
- 多Agent工作区:运行OpenClaw以及内置Agent(Gemini/OpenAI/Anthropic/Ollama)、Claude Code、Codex等
|
||||
|
||||
## Key Concepts
|
||||
- [[AionUi]]:支持OpenClaw和其他Agent的桌面应用
|
||||
- [[远程救援]]:远程修复OpenClaw的能力
|
||||
- [[Cowork空间]]:Agent的桌面工作区界面
|
||||
|
||||
## Connections
|
||||
- [[AionUi]] ← hosts ← [[OpenClaw]]
|
||||
- [[AionUi]] ← provides ← [[远程救援]]
|
||||
30
wiki/sources/arxiv-paper-reader.md
Normal file
30
wiki/sources/arxiv-paper-reader.md
Normal file
@@ -0,0 +1,30 @@
|
||||
---
|
||||
title: "arXiv Paper Reader"
|
||||
type: source
|
||||
tags: [openclaw, arxiv, research, automation]
|
||||
date: 2026-03-06
|
||||
---
|
||||
|
||||
## Source File
|
||||
- [[raw/Agent/usecases/arxiv-paper-reader.md]]
|
||||
|
||||
## Summary
|
||||
- 核心主题:对话式arXiv论文阅读助手
|
||||
- 问题域:阅读arXiv论文意味着下载PDF,在论文间切换丢失上下文
|
||||
- 方法/机制:安装arxiv-reader skill,获取干净可读的文本,自动扁平化LaTeX
|
||||
- 结论/价值:在工作区内对话式阅读、分析和比较论文
|
||||
|
||||
## Key Claims
|
||||
- 通过ID获取任何arXiv论文,获得干净可读的文本(LaTeX自动扁平化)
|
||||
- 先浏览论文结构——列出章节以决定在提交全文前读什么
|
||||
- 快速扫描多篇论文摘要以分类阅读列表
|
||||
- 要求Agent总结、比较或批评特定章节
|
||||
- 结果在本地缓存——重新访问论文是即时的
|
||||
|
||||
## Key Concepts
|
||||
- [[arXiv论文阅读]]:学术论文自动化阅读
|
||||
- [[LaTeX扁平化]]:将LaTeX源文件转换为可读文本
|
||||
- [[arxiv-reader]]:读取arXiv论文的skill
|
||||
|
||||
## Connections
|
||||
- [[arxiv-reader]] ← fetches ← [[arXiv]]
|
||||
29
wiki/sources/autonomous-game-dev-pipeline.md
Normal file
29
wiki/sources/autonomous-game-dev-pipeline.md
Normal file
@@ -0,0 +1,29 @@
|
||||
---
|
||||
title: "Autonomous Educational Game Development Pipeline"
|
||||
type: source
|
||||
tags: [openclaw, autonomous, game-dev, education]
|
||||
date: 2026-03-06
|
||||
---
|
||||
|
||||
## Source File
|
||||
- [[raw/Agent/usecases/autonomous-game-dev-pipeline.md]]
|
||||
|
||||
## Summary
|
||||
- 核心主题:自主教育游戏开发管道
|
||||
- 问题域:独立开发者需要为孩子创建40+教育游戏
|
||||
- 方法/机制:Game Developer Agent自主管理游戏创建和维护的完整生命周期
|
||||
- 结论/价值:每7分钟生产1个新游戏或错误修复
|
||||
|
||||
## Key Claims
|
||||
- 管道能够每7分钟生产1个新游戏或错误修复
|
||||
- 实施严格遵循game-design-rules.md(无框架、移动优先、离线支持)
|
||||
- "错误优先"政策:Agent必须先检查并解决报告的错误,然后才能实施新功能
|
||||
|
||||
## Key Insights
|
||||
- 游戏开发者Agent将LLM变成尊重项目 rigid structure的纪律开发者
|
||||
- 开发队列管理:循环策略平衡跨年龄组的内容
|
||||
- 自动化部署:处理Git工作流程:获取master、创建feature分支、提交更改、合并
|
||||
|
||||
## Key Concepts
|
||||
- [[自主开发管道]]:自动化游戏开发和部署
|
||||
- [[游戏设计规则]]:游戏
|
||||
35
wiki/sources/autonomous-project-management.md
Normal file
35
wiki/sources/autonomous-project-management.md
Normal file
@@ -0,0 +1,35 @@
|
||||
---
|
||||
title: "Autonomous Project Management with Subagents"
|
||||
type: source
|
||||
tags: [openclaw, subagent, project-management, state]
|
||||
date: 2026-03-06
|
||||
---
|
||||
|
||||
## Source File
|
||||
- [[raw/Agent/usecases/autonomous-project-management.md]]
|
||||
|
||||
## Summary
|
||||
- 核心主题:使用子Agent的分散式项目管理
|
||||
- 问题域:传统orchestrator模式造成瓶颈,主Agent成为交通警察
|
||||
- 方法/机制:Agent通过共享STATE.yaml文件协调,多个子Agent并行工作
|
||||
- 结论/价值:无orchestrator开销,主会话保持精简
|
||||
|
||||
## Key Claims
|
||||
- 分散式协调:Agent读写共享STATE.yaml文件
|
||||
- 并行执行:多个子Agent同时处理独立任务
|
||||
- 无orchestrator开销:主会话仅策略执行
|
||||
- 自文档化:所有任务状态持久化在版本控制文件中
|
||||
|
||||
## Key Insights
|
||||
- STATE.yaml > orchestrator:基于文件的协调比消息传递更具可扩展性
|
||||
- Git作为审计日志:提交STATE.yaml更改以获取完整历史
|
||||
- 标签约定很重要:使用pm-{project}-{scope}便于追踪
|
||||
|
||||
## Key Concepts
|
||||
- [[分散式协调]]:通过共享文件而非中央协调器进行协调
|
||||
- [[STATE.yaml]]:项目协调文件,作为单一事实来源
|
||||
- [[子Agent]]:独立执行任务的Agent
|
||||
|
||||
## Connections
|
||||
- [[子Agent]] ← read_write ← [[STATE.yaml]]
|
||||
- [[主Agent]] ← spawns ← [[子Agent]]
|
||||
37
wiki/sources/best-7-news-api.md
Normal file
37
wiki/sources/best-7-news-api.md
Normal file
@@ -0,0 +1,37 @@
|
||||
---
|
||||
title: "Best 7 News API Data Feeds"
|
||||
type: source
|
||||
tags: [news-api, data-feed, API]
|
||||
date: 2025-03-11
|
||||
---
|
||||
|
||||
## Source File
|
||||
- [[raw/AI/Best 7 news API data feeds - AI News.md]]
|
||||
|
||||
## Summary
|
||||
- 核心主题:7大新闻API数据源评测
|
||||
- 问题域:获取实时和历史新闻数据
|
||||
- 方法/机制:聚合、整理多来源结构化新闻数据
|
||||
- 结论/价值:不同API适合不同场景,金融、媒体、风险评估各有最佳选择
|
||||
|
||||
## Key Claims
|
||||
- Webz.io:最全面,支持开放网、深网、暗网数据
|
||||
- GNews API:轻量级,适合初创企业和小型应用
|
||||
- The Guardian API:高质量编辑内容
|
||||
- Bloomberg API:专注金融市场和投资专业数据
|
||||
- Financial Times API:全球金融和市场深度洞察
|
||||
- Opoint:媒体监测和情感分析
|
||||
- Mediastack:可扩展性强,支持7500+来源
|
||||
|
||||
## Key Concepts
|
||||
- [[新闻API]]:聚合、组织、传递结构化新闻数据的平台
|
||||
- [[实时数据]]:新闻API提供的即时新闻覆盖
|
||||
- [[情感分析]]:通过API数据进行品牌声誉监测
|
||||
|
||||
## Key Entities
|
||||
|
||||
## Connections
|
||||
- [[数据聚合]] ← uses ← [[新闻API]]
|
||||
- [[金融情报]] ← uses ← [[Bloomberg API]]
|
||||
|
||||
## Contradictions
|
||||
33
wiki/sources/build-your-own-x.md
Normal file
33
wiki/sources/build-your-own-x.md
Normal file
@@ -0,0 +1,33 @@
|
||||
---
|
||||
title: "Build Your Own X"
|
||||
type: source
|
||||
tags: [build-your-own-x, learning, programming, tutorials]
|
||||
date: 2026-01-01
|
||||
---
|
||||
|
||||
## Source File
|
||||
- [[raw/AI/codecrafters-iobuild-your-own-x Master programming by recreating your favorite technologies from scratch.md]]
|
||||
|
||||
## Summary
|
||||
- 核心主题:Build Your Own X编程学习资源合集
|
||||
- 问题域:学习编程需要通过实践重建技术来深入理解
|
||||
- 方法/机制:提供分步骤指南,从零开始重建流行技术
|
||||
- 结论/价值:费曼的名言"我不能创造的东西,我就不理解"
|
||||
|
||||
## Key Claims
|
||||
- 涵盖23个技术类别:3D渲染器、AR、BitTorrent、区块链、Bot、CLI工具、数据库、Docker、模拟器、前端框架、游戏、Git、网络栈、神经网络、操作系统、物理引擎、编程语言、正则引擎、搜索引擎、Shell、模板引擎、文本编辑器、视觉识别系统、体素引擎、网页浏览器、网页服务器
|
||||
- 每种技术都提供多种编程语言的教程
|
||||
- 著名引用:Richard Feynman的"我不能创造的东西,我就不理解"
|
||||
|
||||
## Key Concepts
|
||||
- [[Build Your Own X]]:通过重建技术来学习的编程学习法
|
||||
- [[CodeCrafters]]:提供编程挑战的平台
|
||||
|
||||
## Key Entities
|
||||
- [[CodeCrafters]]:编程学习平台
|
||||
- [[Richard Feynman]]:物理学家名言出处
|
||||
|
||||
## Connections
|
||||
- [[编程学习]] ← uses ← [[Build Your Own X]]
|
||||
|
||||
## Contradictions
|
||||
39
wiki/sources/claude-skills.md
Normal file
39
wiki/sources/claude-skills.md
Normal file
@@ -0,0 +1,39 @@
|
||||
---
|
||||
title: "Claude Skills最值得研究的AI范式"
|
||||
type: source
|
||||
tags: [claude-skills, prompt-engineering, workflow, vibe-coding]
|
||||
date: 2026-01-05
|
||||
---
|
||||
|
||||
## Source File
|
||||
- [[raw/AI/3.2 万人收藏的 Claude Skills,才是 AI 这条路上最值得研究的一套范式!.md]]
|
||||
|
||||
## Summary
|
||||
- 核心主题:Claude Skills范式深度解析
|
||||
- 问题域:提示词工程向流程工程的转变需求
|
||||
- 方法/机制:Skills作为AI的"说明书"和"SOP",实现稳定复用和自动执行
|
||||
- 结论/价值:Skills的爆发标志着从提示词工程迈向流程工程
|
||||
|
||||
## Key Claims
|
||||
- Skills是Anthropic官方发布的AI技能指南,本质是写给Claude的"说明书"和"SOP"
|
||||
- 官方Skills仓库展示:办公自动化四大件(Word/PDF/PPT/Excel)、开发者工具箱、创意类Skill
|
||||
- 三大Awesome-Claude-Skills仓库:ComposioHQ、VoltAgent、BehiSecc
|
||||
- Skills聚合站:skillsmp.com、aitmpl.com/skills、claudemarketplaces.com
|
||||
- Skills爆发标志从提示词工程到流程工程的关键转变
|
||||
|
||||
## Key Concepts
|
||||
- [[Claude Skills]]:AI技能的系统化封装,包含Prompt结构、参数含义、容错策略
|
||||
- [[流程工程]]:Skills将经验沉淀为SOP,交给AI稳定执行
|
||||
- [[Vibe Coding]]:AI编程方式,其尽头也是Skills
|
||||
- [[SOP标准化]]:将重复任务拆解为AI能理解、稳定复用的流程
|
||||
|
||||
## Key Entities
|
||||
- [[Anthropic]]:Claude Skills官方仓库发布者
|
||||
- [[Claude]]:AI助手
|
||||
|
||||
## Connections
|
||||
- [[提示词工程]] ← evolves_to ← [[流程工程]]
|
||||
- [[Claude Skills]] ← implements ← [[SOP标准化]]
|
||||
- [[Vibe Coding]] ← uses ← [[Claude Skills]]
|
||||
|
||||
## Contradictions
|
||||
25
wiki/sources/content-factory.md
Normal file
25
wiki/sources/content-factory.md
Normal file
@@ -0,0 +1,25 @@
|
||||
---
|
||||
title: "Multi-Agent Content Factory"
|
||||
type: source
|
||||
tags: [openclaw, multi-agent, content, discord]
|
||||
date: 2026-03-06
|
||||
---
|
||||
|
||||
## Source File
|
||||
- [[raw/Agent/usecases/content-factory.md]]
|
||||
|
||||
## Summary
|
||||
- 核心主题:Discord中的多Agent内容工厂
|
||||
- 问题域:内容创作者需要在研究、写作、设计之间切换,耗时巨大
|
||||
- 方法/机制:研究Agent扫描趋势,写作Agent生成内容,缩略图Agent生成图片
|
||||
- 结论/价值:Agent链式协作实现完全无人值守的内容生产
|
||||
|
||||
## Key Claims
|
||||
- 研究Agent feeds写作Agent,写作Agent feeds缩略图Agent
|
||||
- Discord频道便于分别审查每个Agent的工作
|
||||
- 可适配任何内容格式:推文、新闻通讯、LinkedIn帖子等
|
||||
|
||||
## Key Concepts
|
||||
- [[多Agent协作]]:多个专业Agent链式工作
|
||||
- [[内容工厂]]:自动化内容生产流水线
|
||||
- [[Discord集成]]:通过Discord协调多
|
||||
25
wiki/sources/custom-morning-brief.md
Normal file
25
wiki/sources/custom-morning-brief.md
Normal file
@@ -0,0 +1,25 @@
|
||||
---
|
||||
title: "Custom Morning Brief"
|
||||
type: source
|
||||
tags: [openclaw, automation, morning, telegram]
|
||||
date: 2026-03-06
|
||||
---
|
||||
|
||||
## Source File
|
||||
- [[raw/Agent/usecases/custom-morning-brief.md]]
|
||||
|
||||
## Summary
|
||||
- 核心主题:每日定时发送个性化早间简报
|
||||
- 问题域:早晨花费30分钟获取当天动态,而非工作
|
||||
- 方法/机制:Agent在夜间生成内容,早晨发送结构化简报
|
||||
- 结论/价值:AI推荐任务部分是蕞强大的,让Agent主动思考如何帮助
|
||||
|
||||
## Key Claims
|
||||
- AI推荐任务部分蕞强大:让Agent主动思考如何帮助,而非等待指令
|
||||
- 全量draft(而非仅仅想法)是节省时间的关键
|
||||
- 通过短信自定义简报:说"添加到早间简报"即可
|
||||
|
||||
## Key Insights
|
||||
- 夜间生成内容,早晨醒来即可工作
|
||||
- AI推荐任务让Agent主动思考如何帮助用户
|
||||
- 完全draft而非
|
||||
26
wiki/sources/daily-reddit-digest.md
Normal file
26
wiki/sources/daily-reddit-digest.md
Normal file
@@ -0,0 +1,26 @@
|
||||
---
|
||||
title: "Daily Reddit Digest"
|
||||
type: source
|
||||
tags: [openclaw, reddit, content, automation]
|
||||
date: 2026-03-06
|
||||
---
|
||||
|
||||
## Source File
|
||||
- [[raw/Agent/usecases/daily-reddit-digest.md]]
|
||||
|
||||
## Summary
|
||||
- 核心主题:每日Reddit内容摘要
|
||||
- 问题域:想浏览Reddit但时间有限
|
||||
- 方法/机制:reddit-readonly skill获取子版块热门帖子,每日定时摘要
|
||||
- 结论/价值:按需获取Reddit热门内容,无需手动浏览
|
||||
|
||||
## Key Claims
|
||||
- reddit-readonly skill不需要认证
|
||||
- 每日下午5点运行此流程并给出摘要
|
||||
- 创建单独记忆追踪用户偏好
|
||||
|
||||
## Key Concepts
|
||||
- [[Reddit聚合]]:收集Reddit热门内容
|
||||
- [[内容过滤]]:根据用户偏好过滤内容
|
||||
|
||||
## Connections
|
||||
28
wiki/sources/daily-youtube-digest.md
Normal file
28
wiki/sources/daily-youtube-digest.md
Normal file
@@ -0,0 +1,28 @@
|
||||
---
|
||||
title: "Daily YouTube Digest"
|
||||
type: source
|
||||
tags: [openclaw, youtube, content, automation]
|
||||
date: 2026-03-06
|
||||
---
|
||||
|
||||
## Source File
|
||||
- [[raw/Agent/usecases/daily-youtube-digest.md]]
|
||||
|
||||
## Summary
|
||||
- 核心主题:每日自动获取YouTube频道更新并生成摘要
|
||||
- 问题域:YouTube通知不可靠,优质内容容易被错过
|
||||
- 方法/机制:安装youtube-full skill,自动获取频道最新视频并生成摘要
|
||||
- 结论/价值:每天早晨收到个性化内容摘要,避免算法推荐的信息茧房
|
||||
|
||||
## Key Claims
|
||||
- YouTube通知不可靠,订阅的频道新视频不会出现在通知中
|
||||
- youtube-full skill支持100个免费积分注册
|
||||
- channel/latest和channel/resolve免费(0积分)
|
||||
- 仅转录才需要积分
|
||||
|
||||
## Key Concepts
|
||||
- [[内容聚合]]:从多个来源收集内容
|
||||
- [[视频摘要]]:将视频内容压缩为关键点
|
||||
|
||||
## Connections
|
||||
- [[youtube-full]] ← fetches ← [[YouTube]]
|
||||
35
wiki/sources/designing-for-agentic-ai.md
Normal file
35
wiki/sources/designing-for-agentic-ai.md
Normal file
@@ -0,0 +1,35 @@
|
||||
---
|
||||
title: "Designing for Agentic AI"
|
||||
type: source
|
||||
tags: [agentic-ai, product-design, UX]
|
||||
date: 2025-03-02
|
||||
---
|
||||
|
||||
## Source File
|
||||
- [[raw/AI/Designing for Agentic AI.md]]
|
||||
|
||||
## Summary
|
||||
- 核心主题:Agentic AI产品设计原则
|
||||
- 问题域:AI从被动响应到主动行动的转变
|
||||
- 方法/机制:透明性、控制权、个性化、对话、预判五大设计原则
|
||||
- 结论/价值:Agentic AI需要新的设计范式,强调实时反馈和用户控制
|
||||
|
||||
## Key Claims
|
||||
- GenAI擅长内容生成,Agentic AI擅长行动执行
|
||||
- Agentic AI引入新维度:主动 agent 预判需求并自主行动
|
||||
- 用户通过观察AI决策过程进行交互,而非传统点击输入
|
||||
- 五大设计原则:透明性、控制权、个性化、对话、预判
|
||||
|
||||
## Key Concepts
|
||||
- [[GenAI]]:生成式AI,擅长创作内容
|
||||
- [[Agentic AI]]:智能体AI,能够自主行动和决策
|
||||
- [[实时反馈]]:Agentic AI设计中的核心用户体验要素
|
||||
- [[用户控制]]:确保用户对AI行为有最终决定权
|
||||
|
||||
## Key Entities
|
||||
|
||||
## Connections
|
||||
- [[Agentic AI]] ← extends ← [[GenAI]]
|
||||
- [[AI产品设计]] ← uses ← [[Agentic AI设计原则]]
|
||||
|
||||
## Contradictions
|
||||
28
wiki/sources/dynamic-dashboard.md
Normal file
28
wiki/sources/dynamic-dashboard.md
Normal file
@@ -0,0 +1,28 @@
|
||||
---
|
||||
title: "Dynamic Dashboard with Sub-agent Spawning"
|
||||
type: source
|
||||
tags: [openclaw, dashboard, subagent, monitoring]
|
||||
date: 2026-03-06
|
||||
---
|
||||
|
||||
## Source File
|
||||
- [[raw/Agent/usecases/dynamic-dashboard.md]]
|
||||
|
||||
## Summary
|
||||
- 核心主题:动态仪表板通过子Agent并行获取数据
|
||||
- 问题域:静态仪表板显示过时数据,需要手动更新
|
||||
- 方法/机制:对话式定义监控目标,子Agent并行获取每个数据源
|
||||
- 结论/价值:实时可视化多个数据源,无需构建自定义前端
|
||||
|
||||
## Key Claims
|
||||
- 并行获取数据避免阻塞和分配API负载
|
||||
- 聚合结果到统一仪表板(文本、HTML或Canvas)
|
||||
- 指标存储在数据库中用于历史分析
|
||||
- 指标跨阈值时发送警报
|
||||
|
||||
## Key Concepts
|
||||
- [[动态仪表板]]:实时更新的可视化面板
|
||||
- [[并行处理]]:同时获取多个数据源
|
||||
- [[子Agent]]:用于并行执行任务的Agent
|
||||
|
||||
## Connections
|
||||
27
wiki/sources/earnings-tracker.md
Normal file
27
wiki/sources/earnings-tracker.md
Normal file
@@ -0,0 +1,27 @@
|
||||
---
|
||||
title: "AI-Powered Earnings Tracker"
|
||||
type: source
|
||||
tags: [openclaw, finance, tracking, telegram]
|
||||
date: 2026-03-06
|
||||
---
|
||||
|
||||
## Source File
|
||||
- [[raw/Agent/usecases/earnings-tracker.md]]
|
||||
|
||||
## Summary
|
||||
- 核心主题:自动化追踪科技公司财报
|
||||
- 问题域:跟踪数十家科技公司财报需要多来源检查
|
||||
- 方法/机制:每周扫描财报日历,定时任务获取结果并发送摘要
|
||||
- 结论/价值:自动化财报追踪和传递
|
||||
|
||||
## Key Claims
|
||||
- 每周日预览:扫描即将到来周的财报日历
|
||||
- 用户选择要追踪的公司,Agent为每个财报日期安排一次性cron任务
|
||||
- 报告发布后自动搜索结果,格式化摘要并发送
|
||||
|
||||
## Key Concepts
|
||||
- [[财报追踪]]:监控公司财务报告
|
||||
- [[定时任务]]:按计划执行自动化任务
|
||||
|
||||
## Connections
|
||||
- [[财报追踪]] ← uses ← [[定时任务]]
|
||||
33
wiki/sources/event-guest-confirmation.md
Normal file
33
wiki/sources/event-guest-confirmation.md
Normal file
@@ -0,0 +1,33 @@
|
||||
---
|
||||
title: "Event Guest Confirmation"
|
||||
type: source
|
||||
tags: [openclaw, phone, automation, supercall]
|
||||
date: 2026-03-06
|
||||
---
|
||||
|
||||
## Source File
|
||||
- [[raw/Agent/usecases/event-guest-confirmation.md]]
|
||||
|
||||
## Summary
|
||||
- 核心主题:使用SuperCall自动确认活动嘉宾出席
|
||||
- 问题域:手动电话确认20+客人繁琐且容易遗漏
|
||||
- 方法/机制:AI逐个呼叫客人,确认出席并收集备注,编译摘要
|
||||
- 结论/价值:真人电话获得更高响应率
|
||||
|
||||
## Key Claims
|
||||
- 真人电话获得比短信更高的响应率
|
||||
- SuperCall是完全独立的语音Agent,只能访问提供的上下文
|
||||
- 每次通话后AI persona重置,避免对话间交叉污染
|
||||
|
||||
## Key Insights
|
||||
- 从小规模测试开始:用2-3个客人测试persona和开场白
|
||||
- 注意拨打电话时间:不要在太早或太晚打电话
|
||||
- 审核转录:首次批量通话后浏览对话进展
|
||||
|
||||
## Key Concepts
|
||||
- [[电话确认]]:通过电话确认出席
|
||||
- [[SuperCall]]:独立语音Agent工具
|
||||
- [[批量呼叫]]:自动逐个呼叫列表中的联系人
|
||||
|
||||
## Connections
|
||||
- [[SuperCall]] ← used_for ← [[电话确认]]
|
||||
33
wiki/sources/family-calendar-household-assistant.md
Normal file
33
wiki/sources/family-calendar-household-assistant.md
Normal file
@@ -0,0 +1,33 @@
|
||||
---
|
||||
title: "Family Calendar Aggregation & Household Assistant"
|
||||
type: source
|
||||
tags: [openclaw, family, calendar, automation]
|
||||
date: 2026-03-06
|
||||
---
|
||||
|
||||
## Source File
|
||||
- [[raw/Agent/usecases/family-calendar-household-assistant.md]]
|
||||
|
||||
## Summary
|
||||
- 核心主题:家庭日历聚合和家务助理
|
||||
- 问题域:现代家庭跨越多个平台和格式处理五个或更多日历
|
||||
- 方法/机制:OpenClaw作为常开的家务协调员,聚合日历并监控消息
|
||||
- 结论/价值:早上简报包含所有家庭日历的单一视图
|
||||
|
||||
## Key Claims
|
||||
- 日历聚合:从所有家庭日历来源编译成单一每日简报
|
||||
- 环境消息监控:被动监控并在检测到约会时创建日历事件
|
||||
- 驾驶时间缓冲:添加旅行时间块在检测到的约会前后
|
||||
- 家庭库存:维护 pantry/冰箱物品的运行库存
|
||||
|
||||
## Key Insights
|
||||
- 环境 > 主动:蕞大的突破是Agent无需询问即可行动
|
||||
- Mac Mini是甜区:家庭Mac Mini运行受益最大——iMessage集成、Apple Calendar、常开可用性
|
||||
- 从只读开始:在启用写操作之前从日历读取和消息监控开始
|
||||
|
||||
## Key Concepts
|
||||
- [[日历聚合]]:整合多个日历来源
|
||||
- [[环境监控]]:被动监控和自动操作
|
||||
- [[家庭协调]]:管理家庭物流和沟通
|
||||
|
||||
## Connections
|
||||
37
wiki/sources/gemini-product-manager-prd.md
Normal file
37
wiki/sources/gemini-product-manager-prd.md
Normal file
@@ -0,0 +1,37 @@
|
||||
---
|
||||
title: "不会Gemini的产品经理真的要被淘汰了"
|
||||
type: source
|
||||
tags: [gemini, 产品经理, PRD, AI工作流]
|
||||
date: 2025-11-19
|
||||
---
|
||||
|
||||
## Source File
|
||||
- [[raw/AI/不会Gemini的产品经理真的要被淘汰了 附保姆级PRD生成指南.md]]
|
||||
|
||||
## Summary
|
||||
- 核心主题:AI时代产品经理的工作方式变革
|
||||
- 问题域:产品经理如何利用Gemini提效90%以上
|
||||
- 方法/机制:FeatureList构思→逻辑图→PRD文档→HTML原型
|
||||
- 结论/价值:AI是工具,关键在于产品经理的市场洞察能力
|
||||
|
||||
## Key Claims
|
||||
- Gemini 2.5/3可将产品经理文本工作时间缩短90%以上
|
||||
- 核心方法:LLM负责"写"而非"想",人类负责"想"
|
||||
- FeatureList分层级展开功能点:模块分类→功能点全面性→优先级
|
||||
- 用mermaid生成ER图、时序图、甘特图等逻辑图
|
||||
- Gemini可以生成HTML替代原型图
|
||||
|
||||
## Key Concepts
|
||||
- [[FeatureList]]:分层级需求表
|
||||
- [[PRD]]:产品需求文档
|
||||
- [[mermaid]]:图表描述语言
|
||||
- [[AI工作流]]:人类思考+AI执行的协作模式
|
||||
|
||||
## Key Entities
|
||||
- [[Gemini]]:Google AI模型
|
||||
|
||||
## Connections
|
||||
- [[产品经理]] ← uses ← [[Gemini]]
|
||||
- [[PRD生成]] ← uses ← [[AI工作流]]
|
||||
|
||||
## Contradictions
|
||||
36
wiki/sources/gu-ding-jing-tou-duan-shi-pin-ai-sheng-chan.md
Normal file
36
wiki/sources/gu-ding-jing-tou-duan-shi-pin-ai-sheng-chan.md
Normal file
@@ -0,0 +1,36 @@
|
||||
---
|
||||
title: "固定镜头短视频AI全流程制作"
|
||||
type: source
|
||||
tags: [AI视频, 短视频制作, 家装视频]
|
||||
date: 2025-03-15
|
||||
---
|
||||
|
||||
## Source File
|
||||
- [[raw/AI/固定镜头短视频制作的AI全流程解析.md]]
|
||||
|
||||
## Summary
|
||||
- 核心主题:固定镜头短视频AI制作全流程
|
||||
- 问题域:家装类短视频制作效率低
|
||||
- 方法/机制:分镜拆解→九宫格图片生成→首尾针动画→剪辑配音
|
||||
- 结论/价值:AI可将10分钟制作周期缩短至极限
|
||||
|
||||
## Key Claims
|
||||
- 三大关键词:固定机位、内容连续变化、时间压缩
|
||||
- AI工具分类:大脑类(XAR GPT)、设计师类(Midjourney/Nano Banana)、动效类(海螺AI/KAI)
|
||||
- 九宫格法保证画面一致性
|
||||
- 首尾针动画实现平滑过渡
|
||||
- 五步公式:拆分镜头→一致性图像→首尾针动画→快速剪辑→声音设计
|
||||
|
||||
## Key Concepts
|
||||
- [[固定机位]]:摄像机位置固定不变
|
||||
- [[首尾针动画]]:通过首尾帧AI自动补齐中间动作
|
||||
- [[九宫格法]]:一次性生成3x3共九个分镜画面
|
||||
- [[时间压缩]]:将长时间过程浓缩呈现
|
||||
|
||||
## Key Entities
|
||||
|
||||
## Connections
|
||||
- [[AI视频制作]] ← uses ← [[首尾针动画]]
|
||||
- [[AI视频制作]] ← uses ← [[九宫格法]]
|
||||
|
||||
## Contradictions
|
||||
34
wiki/sources/habit-tracker-accountability-coach.md
Normal file
34
wiki/sources/habit-tracker-accountability-coach.md
Normal file
@@ -0,0 +1,34 @@
|
||||
---
|
||||
title: "Habit Tracker & Accountability Coach"
|
||||
type: source
|
||||
tags: [openclaw, habit, tracking, telegram]
|
||||
date: 2026-03-06
|
||||
---
|
||||
|
||||
## Source File
|
||||
- [[raw/Agent/usecases/habit-tracker-accountability-coach.md]]
|
||||
|
||||
## Summary
|
||||
- 核心主题:主动问责Partner追踪习惯
|
||||
- 问题域:习惯追踪App依赖用户记得打开,被动
|
||||
- 方法/机制:Agent通过Telegram或SMS主动联系,根据表现调整语气
|
||||
- 结论/价值:主动问责比被动追踪更有效
|
||||
|
||||
## Key Claims
|
||||
- 真正有效的是主动问责
|
||||
- 每日签到根据选择的时间通过Telegram或SMS发送
|
||||
- 追踪用户定义的习惯
|
||||
- 连续追踪:知道每个习惯的当前连续天数并在消息中提及
|
||||
|
||||
## Key Insights
|
||||
- 适应性语气让这不同于cron job
|
||||
- 保持追踪的习惯数量少(3-5个)
|
||||
- 每周模式分析出奇地有用
|
||||
|
||||
## Key Concepts
|
||||
- [[习惯追踪]]:长期追踪行为习惯
|
||||
- [[主动问责]]:Agent主动联系用户进行检查
|
||||
- [[适应性语气]]:根据表现调整沟通风格
|
||||
|
||||
## Connections
|
||||
- [[习惯追踪]] ← uses ← [[Telegram]]
|
||||
28
wiki/sources/health-symptom-tracker.md
Normal file
28
wiki/sources/health-symptom-tracker.md
Normal file
@@ -0,0 +1,28 @@
|
||||
---
|
||||
title: "Health & Symptom Tracker"
|
||||
type: source
|
||||
tags: [openclaw, health, tracking, telegram]
|
||||
date: 2026-03-06
|
||||
---
|
||||
|
||||
## Source File
|
||||
- [[raw/Agent/usecases/health-symptom-tracker.md]]
|
||||
|
||||
## Summary
|
||||
- 核心主题:追踪食物和症状以识别敏感源
|
||||
- 问题域:识别食物敏感性需要长期持续记录
|
||||
- 方法/机制:通过Telegram主题记录,3次每日提醒,每周分析模式
|
||||
- 结论/价值:自动化追踪和分析帮助识别潜在诱因
|
||||
|
||||
## Key Claims
|
||||
- 识别食物敏感性需要跨时间的持续记录
|
||||
- 每日3次提醒促使记录保持
|
||||
- 每周分析识别与症状相关的食物模式
|
||||
|
||||
## Key Concepts
|
||||
- [[健康追踪]]:长期记录健康数据
|
||||
- [[模式识别]]:从数据中发现规律
|
||||
- [[Telegram集成]]:通过Telegram进行交互
|
||||
|
||||
## Connections
|
||||
- [[健康追踪]] ← uses ← [[Telegram集成]]
|
||||
24
wiki/sources/inbox-declutter.md
Normal file
24
wiki/sources/inbox-declutter.md
Normal file
@@ -0,0 +1,24 @@
|
||||
---
|
||||
title: "Inbox De-clutter"
|
||||
type: source
|
||||
tags: [openclaw, email, automation, gmail]
|
||||
date: 2026-03-06
|
||||
---
|
||||
|
||||
## Source File
|
||||
- [[raw/Agent/usecases/inbox-declutter.md]]
|
||||
|
||||
## Summary
|
||||
- 核心主题:自动化新闻邮件分类和摘要
|
||||
- 问题域:新闻邮件堆积如山,从未打开
|
||||
- 方法/机制:创建专用OpenClaw邮件,订阅新闻,cron job每日摘要
|
||||
- 结论/价值:自动化邮件管理,减少收件箱混乱
|
||||
|
||||
## Key Claims
|
||||
- 为OpenClaw创建专用邮件解决新闻邮件堆积问题
|
||||
- 每日cron job读取过去24小时新闻邮件,生成摘要
|
||||
- 根据反馈更新记忆以更好选择内容
|
||||
|
||||
## Key Concepts
|
||||
- [[邮件管理]]:自动化邮件处理
|
||||
- [[新闻聚合]]:收集和汇总新闻内容
|
||||
22
wiki/sources/knowledge-base-rag.md
Normal file
22
wiki/sources/knowledge-base-rag.md
Normal file
@@ -0,0 +1,22 @@
|
||||
---
|
||||
title: "Knowledge Base RAG"
|
||||
type: source
|
||||
tags: [openclaw, rag, knowledge, automation]
|
||||
date: 2026-03-06
|
||||
---
|
||||
|
||||
## Source File
|
||||
- [[raw/Agent/usecases/knowledge-base-rag.md]]
|
||||
|
||||
## Summary
|
||||
- 核心主题:基于RAG的知识库系统
|
||||
- 问题域:分散的知识难以搜索和利用
|
||||
- 方法/机制:RAG(检索增强生成)系统从知识库中检索相关信息
|
||||
- 结论/价值:让Agent能够基于自有知识库回答问题
|
||||
|
||||
## Key Claims
|
||||
- RAG系统从自有知识库中检索相关信息
|
||||
- 向量数据库支持语义搜索
|
||||
- 定期更新知识库内容
|
||||
|
||||
## Key Concepts
|
||||
Some files were not shown because too many files have changed in this diff Show More
Reference in New Issue
Block a user