Auto-sync: 2026-04-17 22:17
This commit is contained in:
31
wiki/concepts/Ambient-Mode.md
Normal file
31
wiki/concepts/Ambient-Mode.md
Normal file
@@ -0,0 +1,31 @@
|
||||
---
|
||||
id: ambient-mode
|
||||
title: "Ambient Mode(环境模式)"
|
||||
type: concept
|
||||
tags: []
|
||||
aliases:
|
||||
- Ambient
|
||||
last_updated: 2026-04-17
|
||||
---
|
||||
|
||||
## Definition
|
||||
AI Agent 在后台被动监控环境(消息、日历、文件系统等),当检测到可执行事项时主动采取行动,无需用户主动请求。
|
||||
|
||||
## Why It Matters
|
||||
- 传统 AI 交互是"请求-响应"模式,用户必须明确告诉 AI 要做什么
|
||||
- Ambient Mode 实现"感知-行动"模式,AI 主动识别用户需求并提前处理
|
||||
- 本用例中:AI 监控 iMessage,发现牙医预约确认短信后自动创建日历事件并添加驾驶时间缓冲
|
||||
|
||||
## Use Cases
|
||||
- 日历自动创建(从消息中识别预约)
|
||||
- 待办事项提取(从邮件/对话中识别承诺)
|
||||
- 库存变化检测(从购物小票照片更新库存)
|
||||
|
||||
## Related Concepts
|
||||
- [[Cron-Jobs]]:定时任务调度
|
||||
- [[上下文记忆]]:保留对话历史
|
||||
- [[Preference-Learning]]:学习用户偏好
|
||||
|
||||
## Related Entities
|
||||
- [[OpenClaw]]:支持 Ambient Mode 的 AI Agent 平台
|
||||
- [[Mac-Mini]]:适合长期运行的设备
|
||||
20
wiki/concepts/BM25.md
Normal file
20
wiki/concepts/BM25.md
Normal file
@@ -0,0 +1,20 @@
|
||||
---
|
||||
title: "BM25"
|
||||
type: concept
|
||||
tags: []
|
||||
sources: []
|
||||
last_updated: 2026-04-17
|
||||
---
|
||||
|
||||
## Definition
|
||||
BM25(Best Matching 25)是一种基于关键词的全文搜索算法,用于信息检索。
|
||||
|
||||
## Principle
|
||||
基于词频(TF)和逆文档频率(IDF)计算关键词与文档的相关度评分。
|
||||
|
||||
## Use Case
|
||||
- 与向量嵌入结合实现混合搜索,通过 RRF reranking 合并结果
|
||||
|
||||
## Aliases
|
||||
- BM25
|
||||
- Okapi BM25
|
||||
17
wiki/concepts/Game-Developer-Agent.md
Normal file
17
wiki/concepts/Game-Developer-Agent.md
Normal file
@@ -0,0 +1,17 @@
|
||||
---
|
||||
title: "Game Developer Agent"
|
||||
type: concept
|
||||
tags: [AI Agent, Game Development, Automation]
|
||||
---
|
||||
|
||||
## Definition
|
||||
自主管理游戏全生命周期的 AI Agent,负责游戏开发、维护和部署的完整工作流。
|
||||
|
||||
## Related To
|
||||
- [[autonomous-game-dev-pipeline]] — 使用此 Agent 的源案例
|
||||
- [[Bugs First 策略]] — 工作流核心规则
|
||||
- [[Round Robin 策略]] — 任务调度算法
|
||||
|
||||
## Aliases
|
||||
- Game Dev Agent
|
||||
- 游戏开发 Agent
|
||||
26
wiki/concepts/OpenClaw-部署专家.md
Normal file
26
wiki/concepts/OpenClaw-部署专家.md
Normal file
@@ -0,0 +1,26 @@
|
||||
---
|
||||
title: "OpenClaw 部署专家"
|
||||
type: concept
|
||||
tags: [openclaw, deployment, troubleshooting]
|
||||
last_updated: 2026-04-17
|
||||
---
|
||||
|
||||
## Definition
|
||||
AionUi 内置的 OpenClaw 安装、诊断和修复助手,帮助用户完成 OpenClaw 的初始安装、网关配置、故障排查和远程恢复。
|
||||
|
||||
## Components
|
||||
- 安装引导:指导用户完成 OpenClaw 的 npm 安装
|
||||
- 诊断功能:运行 openclaw doctor 检测配置问题
|
||||
- 修复能力:自动修复配置文件、重启网关
|
||||
- 远程救援:通过 WebUI/Telegram 远程执行诊断和修复
|
||||
|
||||
## Use Cases
|
||||
- 新用户首次安装 OpenClaw
|
||||
- OpenClaw 连接失败时的故障排查
|
||||
- 无头运行(headless)或远程机器上的 OpenClaw 恢复
|
||||
- 配置文件的自动修复
|
||||
|
||||
## Related
|
||||
- [[AionUi]] — 部署专家的载体应用
|
||||
- [[OpenClaw]] — 被部署和管理的 AI Agent
|
||||
- [[Remote Rescue]] — 通过远程渠道执行的救援能力
|
||||
20
wiki/concepts/RRF-Reranking.md
Normal file
20
wiki/concepts/RRF-Reranking.md
Normal file
@@ -0,0 +1,20 @@
|
||||
---
|
||||
title: "RRF(Reranking)"
|
||||
type: concept
|
||||
tags: []
|
||||
sources: [https://github.com/zilliztech/memsearch]
|
||||
last_updated: 2026-04-17
|
||||
---
|
||||
|
||||
## Definition
|
||||
RRF(Reciprocal Rank Fusion,倒数排名融合)是一种将多路搜索结果合并排序的算法。
|
||||
|
||||
## Principle
|
||||
对不同搜索方法的结果按排名倒数(1/(k+rank))加权求和,得到综合排名。
|
||||
|
||||
## Use Case
|
||||
- 合并语义搜索(向量嵌入)和关键词搜索(BM25)的结果,提升搜索质量
|
||||
|
||||
## Aliases
|
||||
- RRF
|
||||
- Reciprocal Rank Fusion
|
||||
16
wiki/concepts/Round-Robin-策略.md
Normal file
16
wiki/concepts/Round-Robin-策略.md
Normal file
@@ -0,0 +1,16 @@
|
||||
---
|
||||
title: "Round Robin 策略"
|
||||
type: concept
|
||||
tags: [AI Agent, Task Scheduling]
|
||||
---
|
||||
|
||||
## Definition
|
||||
任务调度算法,轮转分配任务以平衡不同类型/年龄段内容的工作流策略。
|
||||
|
||||
## Related To
|
||||
- [[autonomous-game-dev-pipeline]] — 使用此策略的源案例
|
||||
- [[Game Developer Agent]] — 使用此策略的 Agent
|
||||
|
||||
## Aliases
|
||||
- 轮转调度
|
||||
- 轮询策略
|
||||
32
wiki/concepts/TweetClaw.md
Normal file
32
wiki/concepts/TweetClaw.md
Normal file
@@ -0,0 +1,32 @@
|
||||
---
|
||||
title: "TweetClaw"
|
||||
type: concept
|
||||
tags: [ai-agent, automation, social-media]
|
||||
last_updated: 2026-04-17
|
||||
---
|
||||
|
||||
## Definition
|
||||
TweetClaw 是 OpenClaw 插件,通过自然语言实现 X/Twitter 运营自动化的工具。
|
||||
|
||||
## Related Features
|
||||
- 发推与互动:发推文、回复、点赞、转发、关注/取消关注、发送私信
|
||||
- 搜索与提取:搜索推文和用户,提取粉丝、点赞者、转发者、引用推文者、列表成员
|
||||
- 抽奖管理:从推文互动者中随机抽取获奖者,支持按粉丝数、账号年龄、关键词等条件筛选
|
||||
- 账户监控:监控指定账户的新推文或粉丝变化,主动通知用户
|
||||
|
||||
## Technical Details
|
||||
- 安装方式:`openclaw plugins install @xquik/tweetclaw`
|
||||
- npm 包:`@xquik/tweetclaw`
|
||||
- GitHub 仓库:https://github.com/Xquik-dev/tweetclaw
|
||||
- 所有操作通过托管 API 完成,无需浏览器 Cookie、无需爬虫、无凭证暴露风险
|
||||
|
||||
## Aliases
|
||||
- TweetClaw
|
||||
- X/Twitter Automation
|
||||
|
||||
## Related Concepts
|
||||
- [[工作流自动化]] — 预定义自动化流程
|
||||
- [[自然语言处理]] — 通过自然语言指令驱动操作
|
||||
|
||||
## Related Entities
|
||||
- [[OpenClaw]] — 插件宿主工具
|
||||
23
wiki/concepts/arxiv-reader-skill.md
Normal file
23
wiki/concepts/arxiv-reader-skill.md
Normal file
@@ -0,0 +1,23 @@
|
||||
---
|
||||
title: "arxiv-reader skill"
|
||||
type: concept
|
||||
tags: [agent, skill, arxiv, research]
|
||||
last_updated: 2025-04-17
|
||||
---
|
||||
|
||||
## Definition
|
||||
Prismer 项目提供的 arXiv 论文读取 skill,包含三个核心工具:arxiv_fetch(获取全文)、arxiv_sections(列出章节)、arxiv_abstract(获取摘要)。
|
||||
|
||||
## Key Features
|
||||
- 直接从 arXiv 下载论文源码,自动解压并扁平化 LaTeX
|
||||
- 按章节浏览,避免一次性加载全文
|
||||
- 批量摘要对比,多论文筛选
|
||||
- 本地缓存,重复访问秒级响应
|
||||
|
||||
## Dependencies
|
||||
- Node.js(无需 Docker 或 Python)
|
||||
|
||||
## Use Cases
|
||||
- 学术论文阅读助手
|
||||
- 研究文献调研
|
||||
- 跨论文对比分析
|
||||
47
wiki/concepts/家庭库存追踪.md
Normal file
47
wiki/concepts/家庭库存追踪.md
Normal file
@@ -0,0 +1,47 @@
|
||||
---
|
||||
id: 家庭库存追踪
|
||||
title: "家庭库存追踪"
|
||||
type: concept
|
||||
tags: []
|
||||
aliases:
|
||||
- Pantry Tracking
|
||||
- Household Inventory
|
||||
last_updated: 2026-04-17
|
||||
---
|
||||
|
||||
## Definition
|
||||
通过 JSON 文件维护家庭食品储藏室和冰箱物品的持续库存,支持文本、照片和收据三种更新方式。
|
||||
|
||||
## Implementation
|
||||
```json
|
||||
// ~/household/inventory.json
|
||||
{
|
||||
"items": [
|
||||
{"name": "牛奶", "quantity": 2, "unit": "加仑", "location": "冰箱", "updated": "2026-04-17"},
|
||||
{"name": "鸡蛋", "quantity": 12, "unit": "个", "location": "冰箱", "updated": "2026-04-16"}
|
||||
],
|
||||
"low_stock_threshold": {"牛奶": 1, "鸡蛋": 6}
|
||||
}
|
||||
```
|
||||
|
||||
## Update Methods
|
||||
- **照片输入**:用户发送冰箱/储藏室照片 → 视觉模型提取物品
|
||||
- **文本更新**:"我们没有鸡蛋了" / "买了 2 加仑牛奶"
|
||||
- **收据照片**:购物小票照片 → 自动更新库存
|
||||
|
||||
## Query Capabilities
|
||||
- "我们有黄油吗?" → 返回位置和数量
|
||||
- "什么东西快用完了?" → 列出低于阈值的物品
|
||||
- "生成购物清单" → 汇总低库存物品 + 食谱所需食材
|
||||
|
||||
## Use Cases
|
||||
- 避免重复购买
|
||||
- 减少食物浪费
|
||||
- 简化每周采购
|
||||
|
||||
## Related Concepts
|
||||
- [[OCR]]:光学字符识别
|
||||
- [[Agentic-AI]]:主动行动的 AI 代理
|
||||
|
||||
## Related Entities
|
||||
- [[Airtable]]:可替代 JSON 的在线数据库选项
|
||||
25
wiki/concepts/数据管道-Data-Pipeline.md
Normal file
25
wiki/concepts/数据管道-Data-Pipeline.md
Normal file
@@ -0,0 +1,25 @@
|
||||
---
|
||||
title: "数据管道 (Data Pipeline)"
|
||||
type: concept
|
||||
tags: [automation, data-engineering]
|
||||
last_updated: 2026-04-17
|
||||
---
|
||||
|
||||
## Definition
|
||||
自动化数据采集、清洗、转换和传递的工作流架构,将来自多个源头的数据按预定规则处理后输出到目标系统。
|
||||
|
||||
## Components
|
||||
1. **采集层 (Collection)** — 从 RSS、API、Web 抓取等源头获取原始数据
|
||||
2. **清洗层 (Cleaning)** — 过滤无效数据、规范化格式
|
||||
3. **处理层 (Processing)** — 去重、评分、分类、聚合
|
||||
4. **输出层 (Output)** — 推送至 Discord、邮件、Telegram 等目标
|
||||
|
||||
## Use Cases
|
||||
- [[Multi-Source Tech News Digest]] — 四层数据管道聚合科技新闻
|
||||
- [[Daily Reddit Digest]] — 定时从 Reddit 获取热门帖子
|
||||
- [[电商数据采集系统]] — 多平台商品数据聚合
|
||||
|
||||
## Related Concepts
|
||||
- [[Cron Jobs]] — 定时触发数据管道执行
|
||||
- [[工作流自动化]] — 预定义的自动化处理流程
|
||||
- [[去重算法]] — 数据管道中的重复内容过滤
|
||||
23
wiki/concepts/语义搜索.md
Normal file
23
wiki/concepts/语义搜索.md
Normal file
@@ -0,0 +1,23 @@
|
||||
---
|
||||
title: "语义搜索"
|
||||
type: concept
|
||||
tags: []
|
||||
sources: [https://github.com/zilliztech/memsearch]
|
||||
last_updated: 2026-04-17
|
||||
---
|
||||
|
||||
## Definition
|
||||
语义搜索是通过向量嵌入实现按语义相似度而非关键词匹配的信息检索方式。
|
||||
|
||||
## Principle
|
||||
将文本转换为数值向量(向量嵌入),通过计算向量间的余弦相似度找到语义相近的内容。
|
||||
|
||||
## Use Case
|
||||
- "我们选了什么缓存方案?"能找到包含"cache solution"相关内容,即使问题中不含"caching"一词
|
||||
|
||||
## Relation
|
||||
- 基于 [[向量嵌入]] 技术实现
|
||||
- 结合 [[BM25]] 关键词搜索提升准确率
|
||||
|
||||
## Aliases
|
||||
- Semantic Search
|
||||
23
wiki/concepts/质量评分-Quality-Scoring.md
Normal file
23
wiki/concepts/质量评分-Quality-Scoring.md
Normal file
@@ -0,0 +1,23 @@
|
||||
---
|
||||
title: "质量评分 (Quality Scoring)"
|
||||
type: concept
|
||||
tags: [ai, data-processing, ranking]
|
||||
last_updated: 2026-04-17
|
||||
---
|
||||
|
||||
## Definition
|
||||
通过多维度算法评估内容重要性并赋予分值的机制,用于筛选和排序信息。
|
||||
|
||||
## Scoring Factors
|
||||
- **优先级来源 (Priority Source)**:权威来源 +3 分
|
||||
- **多来源 (Multi-Source)**:多平台报道 +5 分
|
||||
- **时效性 (Recency)**:近期发布 +2 分
|
||||
- **参与度 (Engagement)**:高互动量 +1 分
|
||||
|
||||
## Use Cases
|
||||
- [[Multi-Source Tech News Digest]] — 使用四维度评分筛选科技新闻
|
||||
- [[动态仪表盘]] — 优先显示高权重数据源
|
||||
|
||||
## Related Concepts
|
||||
- [[数据管道]] — 质量评分在数据处理流程中的应用
|
||||
- [[偏好学习]] — 基于用户反馈持续优化评分模型
|
||||
Reference in New Issue
Block a user