Bible Network
Crypto
DeFi
Onchain
RWA
AI Agent
Stablecoin
Chain
SAFU
CryptoTax
DeFAI
AGI
Claude Me
Claude Skill
Claude Design
Claude Cowork
獨立知識媒體
與任何項目無關聯
Claude
Me
繁中
EN
日本語
探索AI智慧的思維邊界
claude-me.com
最新動態
AI 基礎原理
名詞解析
新手入門
工具百科
實戰應用
工具評測
MCP 生態系
最新
Anthropic 親口說的八個原則:Claude 愈用愈笨,很可能是你的方式有問題
·
Claude Code 現在可以自己「循環工作」了:Anthropic 公開四種 Loop 模式,讓 AI 真的幫你跑完整個流程
·
Claude Cowork 真實評測:三個月後,哪些任務真的省了時間,哪些讓我後悔開了自動化
·
七月開局的兩個大消息:Claude Sonnet 5 正式發布,Fable 5 出口管制解除全球重新開放
·
RAG 是什麼:為什麼 Claude 讀不了你公司內網的資料,以及怎麼讓它讀到
·
Claude Cowork 進階工作流:從「交代一件事」到「讓它跑完一整個流程」,三個真實可用的範本
首頁
›
名詞解析
名詞解析 · Claude Me 詞庫
每個術語,都能說清楚
從基礎概念到進階技術,所有你需要理解的術語,按主題分類整理,附生活化比喻與實際例子。
詞條總數
63
主題分類
12
本週新增
1
最近新增
Alignment
AI 對齊
Claude Desktop
Claude Desktop
Prompt Compression
提示詞壓縮
MCP Transport
MCP 傳輸協定
Extended Thinking
延伸思考
Deceptive Alignment
欺騙性對齊
Agent Orchestration
Agent 編排
Claude Cowork
Claude Cowork
Knowledge Graph
知識圖譜
Claude Fable 5
Claude Fable 5
Message Batches API
批次訊息 API
Adaptive Thinking
自適應思考
Global Stablecoin Regulation Landscape
穩定幣監管版圖(Global Stablecoin Regulation Landscape)
Rebase Mechanism
Rebase 機制(彈性供應調節)
Oracle Manipulation
預言機操縱(Oracle Manipulation)
Maker Vault
Maker Vault(造幣金庫)
Stablecoin Bank Run
穩定幣擠兌(Bank Run on Stablecoin)
RLHF (Reinforcement Learning from Human Feedback)
人類回饋強化學習
AI Alignment
AI 對齊
Token
Token(語言模型的基本單位)
Prompt vs System Prompt
提示詞 vs 系統提示詞的差別
MCP (Model Context Protocol)
MCP(模型情境協定)
Claude Model Tiers
Claude 模型梯度
Agentic Workflow
Agent 工作流
Temperature
溫度
Hallucination
幻覺
Few-Shot Prompting
少樣本提示
Claude API
Claude API
Context Window
上下文窗口
Prompt Caching
提示詞快取
搜尋
全部
ai-agents
ai-safety
algorithmic
claude-models
claude-tools
core-concepts
crypto-backed
depegging-risk
mcp-tools
pegging-mechanisms
prompt-techniques
stablecoin-types
ai-agents
共 5 個詞條
看更多 >>
新手
ai-agents
Agentic Loop
Agent 循環
AI Agent 運作的核心模式:感知環境→決策→執行行動→觀察結果→再決策,循環反覆直到任務完成。讓 AI...
中級
ai-agents
Agentic Workflow
Agent 工作流
由 AI Agent 自主執行、包含多個步驟和工具呼叫的任務流程。和一次性的問答不同,Agentic...
中級
ai-agents
Agent Orchestration
Agent 編排
Agent 編排是讓一個「總指揮 Agent」把複雜目標拆解成子任務,再分派給多個「專業...
進階
ai-agents
Multi-Agent Systems
多智能體系統
由多個 AI Agent 協同工作完成複雜任務的架構——不同的 Agent...
中級
ai-agents
Tool Use
工具使用
讓 Claude 能在對話中主動呼叫外部功能的機制——搜尋網路、查詢資料庫、執行程式碼、存取文件系統。開發者在...
ai-safety
共 6 個詞條
看更多 >>
新手
ai-safety
AI Alignment
AI 對齊
讓 AI 系統的目標和行為符合人類的意圖和價值觀的研究與工程工作。簡單說:確保 AI...
新手
ai-safety
AI Alignment
AI 對齊
確保 AI 系統的行為和目標與人類的意圖與價值觀一致的研究領域。簡單說:讓 AI...
新手
ai-safety
AI Safety
AI 安全
確保 AI 系統在開發和部署過程中不會造成意外或有害後果的研究與實踐領域。包含技術層面(讓 AI...
中級
ai-safety
Alignment
AI 對齊
讓 AI 系統的目標、行為與人類真實意圖保持一致的研究領域。對齊失敗時,AI 可能做出「符合指令字面」但違背人類本意的事。
進階
ai-safety
Deceptive Alignment
欺騙性對齊
欺騙性對齊是 <a href="/zh/glossary/ai-safety/ai-safety/">AI...
中級
ai-safety
RLHF (Reinforcement Learning from Human Feedback)
人類回饋強化學習
讓 AI 的行為逐漸符合人類偏好的訓練技術:人工評估者對 AI...
algorithmic
共 1 個詞條
看更多 >>
進階
algorithmic
Rebase Mechanism
Rebase 機制(彈性供應調節)
claude-models
共 5 個詞條
看更多 >>
進階
claude-models
Claude Fable 5
Claude Fable 5
<a...
新手
claude-models
Claude Haiku
Claude...
新手
claude-models
Claude Model Tiers
Claude 模型梯度
<a...
進階
claude-models
Claude Opus
Claude Opus
<a...
中級
claude-models
Claude Sonnet
Claude Sonnet
<a...
claude-tools
共 10 個詞條
看更多 >>
中級
claude-tools
Claude API
Claude API
讓開發者通過程式碼直接呼叫 Claude 能力的介面,可以把 Claude...
新手
claude-tools
Claude API Key
Claude API 金鑰
讓你的應用程式能呼叫 <a...
新手
claude-tools
Claude Artifacts
Claude Artifacts
Claude 在對話旁邊獨立顯示可互動內容的功能——程式碼可以直接執行、網頁可以即時預覽、文件可以單獨編輯。讓...
新手
claude-tools
Claude Code
Claude Code
<a...
新手
claude-tools
Claude Computer Use
Claude 電腦使用
<a...
新手
claude-tools
Claude Cowork
Claude Cowork
<a...
新手
claude-tools
Claude Desktop
Claude Desktop
Claude Desktop 是 <a...
中級
claude-tools
Claude Enterprise
Claude 企業版
<a...
中級
claude-tools
Claude Projects
Claude 專案
Claude.ai 的工作空間功能,讓你建立帶有固定...
中級
claude-tools
Message Batches API
批次訊息 API
<a...
core-concepts
共 19 個詞條
看更多 >>
中級
core-concepts
Adaptive Thinking
自適應思考
<a...
新手
core-concepts
AI Agent
AI 代理人
能自主執行多步驟任務的 AI...
新手
core-concepts
Anthropic
Anthropic
Claude 的開發公司,由前 OpenAI 研究人員於 2021 年創立,核心使命是開發安全且對人類有益的...
中級
core-concepts
Constitutional AI
憲法式 AI
<a...
進階
core-concepts
Context Length Optimization
Context 長度優化
在 API 呼叫中系統性地管理輸入 Token...
新手
core-concepts
Context Window
上下文窗口
AI...
中級
core-concepts
Extended Thinking
延伸思考
延伸思考是 Claude...
中級
core-concepts
Fine-Tuning
微調
在已預訓練的<a...
新手
core-concepts
Hallucination
幻覺
AI...
進階
core-concepts
Inference Optimization
推理優化
降低 AI...
新手
core-concepts
Knowledge Cutoff
知識截止日
AI 模型訓練資料的最後更新時間點,之後發生的事情模型一概不知。Claude 的知識截止日是 2025 年 8...
進階
core-concepts
Knowledge Graph
知識圖譜
用「節點(實體)+...
新手
core-concepts
LLM (Large Language Model)
大型語言模型
以海量文字資料訓練、具備理解和生成人類語言能力的 AI 模型。Claude、GPT-4、Gemini 都是...
進階
core-concepts
Model Distillation
模型蒸餾
用大型「教師」模型的輸出來訓練小型「學生」模型,讓小模型在保留大模型核心能力的同時,大幅降低計算需求。就像讓資深專...
新手
core-concepts
Multimodal
多模態
AI 能夠同時處理和理解多種類型輸入的能力,包含文字、圖片、文件等。Claude...
中級
core-concepts
Prompt Caching
提示詞快取
<a...
中級
core-concepts
Retrieval-Augmented Generation (RAG)
檢索增強生成
讓 AI 在回答前先從外部知識庫檢索相關資料,再根據檢索結果生成回答,解決模型知識過時和幻覺問題的技術架構。
中級
core-concepts
Temperature
溫度
控制 AI 輸出「隨機性」或「創意程度」的參數,通常在 0 到 1 之間。0...
新手
core-concepts
Token
Token(語言模型的基本單位)
AI 語言模型處理文字的最小單位。Token 不等於字——英文大約每 0.75 個詞是一個...
crypto-backed
共 1 個詞條
看更多 >>
中級
crypto-backed
Maker Vault
Maker Vault(造幣金庫)
depegging-risk
共 1 個詞條
看更多 >>
新手
depegging-risk
Stablecoin Bank Run
穩定幣擠兌(Bank Run on Stablecoin)
mcp-tools
共 4 個詞條
看更多 >>
新手
mcp-tools
MCP Client
MCP 客戶端
MCP 架構裡的「使用方」——Claude 本身,以及 <a...
新手
mcp-tools
MCP (Model Context Protocol)
MCP(模型情境協定)
<a...
進階
mcp-tools
MCP Server
MCP 伺服器
MCP(Model Context Protocol)生態系中負責「提供工具和資源」的服務端。開發者用...
新手
mcp-tools
MCP Transport
MCP 傳輸協定
MCP 傳輸協定是 Claude 和 MCP Server...
pegging-mechanisms
共 1 個詞條
看更多 >>
進階
pegging-mechanisms
Oracle Manipulation
預言機操縱(Oracle Manipulation)
prompt-techniques
共 9 個詞條
看更多 >>
進階
prompt-techniques
Chain-of-Thought Prompting
思維鏈提示
讓 AI...
中級
prompt-techniques
Few-Shot Prompting
少樣本提示
在提示詞裡提供幾個「輸入→輸出」的示範例子,讓 AI...
中級
prompt-techniques
Prompt Compression
提示詞壓縮
提示詞壓縮是在不失去關鍵資訊的前提下,縮減送入 AI...
新手
prompt-techniques
Prompt Engineering
提示詞工程
系統化地設計和優化給 AI 的指令,以獲得更準確、更符合需求的輸出。讓同樣的 Claude...
中級
prompt-techniques
Prompt Injection
提示詞注入攻擊
攻擊者透過在外部內容(網頁、文件、使用者輸入)裡嵌入惡意指令,試圖覆蓋或繞過 AI 系統原本的 System...
新手
prompt-techniques
Prompt vs System Prompt
提示詞 vs 系統提示詞的差別
「提示詞」是你在對話裡輸入的問題或指令;「<a...
新手
prompt-techniques
Role Prompting
角色提示
在提示詞開頭指定 Claude...
新手
prompt-techniques
System Prompt
系統提示詞
在對話開始前送給 Claude 的隱藏指令層,用來定義角色、設定規則、限制行為範圍。對話中所有訊息都在...
新手
prompt-techniques
Zero-Shot Prompting
零樣本提示
直接給 AI 一個任務指令,不提供任何範例就期待它完成——「零個示範,直接射出」。和 Few-Shot...
stablecoin-types
共 1 個詞條
看更多 >>
中級
stablecoin-types
Global Stablecoin Regulation Landscape
穩定幣監管版圖(Global Stablecoin Regulation Landscape)
按字母瀏覽
A
B
C
D
E
F
G
H
I
J
K
L
M
N
O
P
Q
R
S
T
U
V
W
X
Y
Z