Design Spec: Bộ Học Liệu "Azure AI Agent"

Tóm tắt dự án

Xây dựng bộ học liệu kỹ thuật chuyên sâu về Azure AI Agent / Microsoft Foundry Agent Service dưới dạng Docs trong Docusaurus, theo chuẩn sư phạm 4MAT, phục vụ đối tượng Intern/Junior Developer muốn học hệ sinh thái Azure AI.

Nguồn tài liệu: sources/documentations/learn.microsoft.com/en-us/azure/foundry/agents/

overview.md (17KB)
concepts/ — 25 files (agent identity, tools, workflows, memory, RAG, MCP, v.v.)
how-to/ — 45+ files (deploy, configure, debug, MCP auth, memory, VS Code, M365, v.v.)
quickstarts/ — 9 files

Quyết định thiết kế

Hạng mục	Quyết định
Output format	`docs/` — Docusaurus Docs với sidebar navigation
Cấu trúc	Tiered Category (nhóm theo chủ đề)
Đối tượng	Intern/Junior Developer
Phong cách	Giảng dạy thuần túy — Concept, Diagram, Analogy
Template	Sử dụng skill`create-tech-lecture` (4MAT: Why → What → How → What if)
Diagrams	Mermaid bắt buộc tại mỗi bước flow/architecture
Glossary	Bảng thuật ngữ đầu mỗi bài
Độ dài	2.000–4.000 từ/bài, 15-20 phút đọc
Footer	"Made by Anh Tu - Share to be share"

Cấu trúc thư mục Docs

docs/
└── azure-ai-agent/
    ├── index.md                              ← Landing page: "Lộ trình học Azure AI Agent"
    │
    ├── 01-foundations/                       ← Category: Nền tảng (Module 1)
    │   ├── index.md
    │   ├── what-is-ai-agent.md
    │   ├── microsoft-foundry.md
    │   ├── development-approaches.md
    │   ├── first-agent-foundry.md
    │   ├── vscode-setup.md
    │   ├── agent-tools-overview.md
    │   └── deploy-integrate.md
    │
    ├── 02-tools/                             ← Category: Công cụ (Module 2, 3)
    │   ├── index.md
    │   ├── custom-tools-why.md               ← Module 2
    │   ├── custom-tools-options.md
    │   ├── custom-tools-integrate.md
    │   ├── mcp-discovery.md                  ← Module 3
    │   ├── mcp-server-client.md
    │   └── mcp-azure-agents.md
    │
    ├── 03-knowledge/                         ← Category: Knowledge & RAG (Module 4)
    │   ├── index.md
    │   ├── rag-for-agents.md
    │   ├── foundry-iq-overview.md
    │   ├── data-sources-config.md
    │   └── retrieval-config.md
    │
    ├── 04-integration/                       ← Category: Tích hợp hệ sinh thái (Module 5, 9)
    │   ├── index.md
    │   ├── m365-publish-options.md           ← Module 5
    │   ├── m365-teams-publish.md
    │   ├── m365-agents-toolkit.md
    │   ├── work-iq.md
    │   ├── a2a-protocol.md                   ← Module 9
    │   ├── a2a-agent-executor.md
    │   ├── a2a-server.md
    │   └── a2a-connect.md
    │
    └── 05-orchestration/                     ← Category: Workflow & Multi-Agent (Module 6, 7, 8)
        ├── index.md
        ├── workflows-overview.md             ← Module 6
        ├── workflow-patterns.md
        ├── workflow-foundry.md
        ├── workflow-power-fx.md
        ├── agent-framework-intro.md          ← Module 7
        ├── agent-framework-create.md
        ├── agent-framework-tools.md
        ├── multi-agent-overview.md           ← Module 8
        ├── orchestration-concurrent.md
        ├── orchestration-sequential.md
        ├── orchestration-group.md
        ├── orchestration-handoff.md
        └── orchestration-magentic.md

Tổng: 5 categories, ~35 files docs

Outline chi tiết & Mapping nguồn tài liệu

🏗️ Category 01: Foundations (Module 1)

`index.md` — Giới thiệu Category

Overview module 1, roadmap trong category

`what-is-ai-agent.md` — AI Agent là gì?

Nguồn: overview.md (lines 7-24) Key concepts:

AI Agent vs Chatbot thông thường
3 core components: Model + Instructions + Tools
Prompt Agent vs Hosted Agent (so sánh bảng)
Analogy: Agent như "nhân viên thông minh" với CV (instructions), kỹ năng (tools), não bộ (model)
Mermaid: Architecture diagram 3 components

`microsoft-foundry.md` — Microsoft Foundry Agent Service là gì?

Nguồn: overview.md (lines 28-34), concepts/runtime-components.md Key concepts:

Foundry = Managed Platform for agents
Responses API — single entry point
Platform components: Runtime, Tools, Models, Observability, Identity
Mermaid: Platform layer diagram
Trade-off: Foundry vs tự host

`development-approaches.md` — Các hướng phát triển Agent

Nguồn: overview.md (lines 36-41), concepts/development-lifecycle.md Key concepts:

3 paths: Prompt Agent (portal-first), Hosted Agent (code-first), Responses API (bring-your-own)
7-step development lifecycle: Create → Test → Trace → Evaluate → Optimize → Publish → Monitor
Mermaid: Decision tree chọn approach, Lifecycle flowchart
Trade-off: Prompt vs Hosted vs BYO

`first-agent-foundry.md` — Build Agent đầu tiên trong Foundry Portal

Nguồn: quickstarts/quickstart-hosted-agent.md, quickstarts/prompt-agent.md Key concepts:

Portal-first flow: Create project → Add model → Configure instructions → Attach tools → Test in playground
Agent playground là gì?
Mermaid: Step-by-step flowchart build agent trong portal

`vscode-setup.md` — Setup VS Code cho Agent Development

Nguồn: how-to/vs-code-agents-workflow-pro-code.md, environment-setup.md Key concepts:

Extensions cần thiết
Pro-code workflow trong VS Code
Connect VS Code với Azure Foundry project
Mermaid: Development environment diagram

`agent-tools-overview.md` — Mở rộng Agent với Tools

Nguồn: concepts/tool-catalog.md (toàn bộ) Key concepts:

Built-in Tools: Web Search, Code Interpreter, File Search, Function Calling
Custom Tools: MCP, OpenAPI, A2A
Toolbox — bundle tools thành 1 endpoint
Authentication cho tools
Mermaid: Tool taxonomy diagram, Tool call flow (sequenceDiagram)
Trade-off: Built-in vs Custom tools

`deploy-integrate.md` — Test, Deploy và Integrate Agent

Nguồn: how-to/agent-applications.md, overview.md (lines 100-125) Key concepts:

Versioning: auto-snapshot, rollback
Publishing: promote to managed resource với stable endpoint
Distribution: Teams, M365 Copilot, Entra Agent Registry
Protocols: OpenResponses, Activity, Invocations, A2A
Mermaid: Deploy pipeline diagram

🔧 Category 02: Tools (Module 2 & 3)

✅ `custom-tools-why.md` — Tại sao cần Custom Tools?

Nguồn: concepts/tool-catalog.md (Custom tools section), concepts/tool-best-practice.md Key concepts:

Gap giữa built-in tools và nhu cầu thực tế doanh nghiệp
Khi nào built-in tools không đủ?
Analogy: Built-in tools như bộ đồ nghề tiêu chuẩn; custom tools như đặt hàng tool riêng
Mermaid: Decision flowchart: Should I use custom tools?

✅ `custom-tools-options.md` — Các lựa chọn Custom Tools

Nguồn: concepts/tool-catalog.md, how-to/tools/function-calling.md, how-to/tools/openapi.md Key concepts:

Function Calling — define & execute locally
OpenAPI Tool — connect REST API bằng spec
MCP — remote tool protocol
A2A — agent-to-agent
Bảng so sánh 4 options theo: hosting, auth, use case, complexity
Mermaid: Comparison diagram

✅ `custom-tools-integrate.md` — Cách tích hợp Custom Tools

Nguồn: how-to/tools/function-calling.md, how-to/tools/openapi.md, how-to/tools/azure-functions.md Key concepts:

Function calling flow: Define schema → Agent calls → App executes → Return result
OpenAPI flow: Provide spec → Agent discovers endpoints → Calls automatically
Mermaid: sequenceDiagram cho từng approach

✅ `mcp-discovery.md` — Understand MCP Tool Discovery

Nguồn: concepts/tool-catalog.md (MCP sections), Tool catalog UI walkthrough Key concepts:

MCP Protocol là gì? (Model Context Protocol)
Remote MCP Server vs Local MCP Server
Tool catalog: browse, filter, configure
Mermaid: MCP topology diagram

✅ `mcp-server-client.md` — MCP Server và Client

Nguồn: how-to/mcp-authentication.md Key concepts:

MCP Server = expose tools via protocol
Authentication: Key-based, Entra (Managed Identity), OAuth OBO
Mermaid: Auth flow diagrams cho từng method
Trade-off: 3 auth methods

✅ `mcp-azure-agents.md` — Dùng MCP với Azure AI Agents

Nguồn: how-to/tools/model-context-protocol.md, overview.md (MCP section) Key concepts:

Thêm MCP tool vào agent definition
Toolbox — bundle MCP tools
Azure Functions as MCP endpoint
Mermaid: End-to-end MCP integration flow

📚 Category 03: Knowledge & RAG (Module 4)

✅ `rag-for-agents.md` — RAG cho AI Agents

Nguồn: concepts/what-is-foundry-iq.md, concepts/vector-stores.md Key concepts:

RAG (Retrieval-Augmented Generation) là gì?
Tại sao cần RAG cho agents? (hallucination problem)
RAG vs Fine-tuning: trade-off quan trọng
Vector Stores trong Foundry
Analogy: Agent như thám tử — RAG cho phép nó tra cứu "hồ sơ vụ án" trước khi trả lời
Mermaid: RAG pipeline diagram

✅ `foundry-iq-overview.md` — Foundry IQ là gì?

Nguồn: concepts/what-is-foundry-iq.md Key concepts:

Foundry IQ = Knowledge base management platform
Foundry IQ vs File Search (so sánh)
Work IQ vs Fabric IQ vs Foundry IQ (bộ 3 IQ)
Mermaid: IQ products overview

✅ `data-sources-config.md` — Cấu hình Data Sources

Nguồn: how-to/foundry-iq-connect.md Key concepts:

Supported data sources: Azure Blob, SharePoint, Azure AI Search, v.v.
Connections trong Foundry project
Data ingestion & chunking strategies
Mermaid: Data source → Index → Knowledge base flow

✅ `retrieval-config.md` — Cấu hình Retrieval với Foundry IQ

Nguồn: how-to/foundry-iq-connect.md, concepts/vector-stores.md Key concepts:

Retrieval config: top-k, score threshold, reranking
Semantic search vs Keyword search
Attach Foundry IQ knowledge base vào agent
Mermaid: Query → Retrieval → Augment → Generate flow

🔗 Category 04: Integration (Module 5 & 9)

`m365-publish-options.md` — Publishing Options trong Foundry

Nguồn: concepts/agent-365-integration.md, overview.md (Publishing section) Key concepts:

4 protocols: OpenResponses, Activity, Invocations, A2A
Entra Agent Registry
Bảng so sánh protocols theo use case
Mermaid: Publishing ecosystem diagram

`m365-teams-publish.md` — Publish Agent lên Microsoft Teams

Nguồn: how-to/publish-copilot.md Key concepts:

Flow publish từ Foundry portal → Teams
M365 Copilot integration
Permissions cần thiết
Mermaid: Publish flow diagram

`m365-agents-toolkit.md` — Microsoft 365 Agents Toolkit (Advanced)

Nguồn: how-to/agent-365.md, how-to/grant-agent-365-permissions.md Key concepts:

Agents Toolkit là gì?
Customization vs Portal publish
Required permissions configuration

`work-iq.md` — Access M365 Data với Work IQ

Nguồn: concepts/what-is-memory.md, how-to/memory-usage.md Key concepts:

Work IQ = M365 data grounding (SharePoint, Teams, Outlook)
Memory vs Work IQ: khi nào dùng cái nào?
Privacy & access control trong M365 data

`a2a-protocol.md` — A2A Protocol là gì?

Nguồn: how-to/enable-agent-to-agent-endpoint.md, concepts/tool-catalog.md (A2A section) Key concepts:

A2A (Agent-to-Agent) Protocol
Khi nào dùng A2A thay vì MCP?
A2A Agent Card concept
Mermaid: A2A communication pattern

`a2a-agent-executor.md` — Implement Agent Executor

Nguồn: how-to/enable-agent-to-agent-endpoint.md Key concepts:

AgentExecutor interface
Task handling & response streaming
Mermaid: Executor lifecycle diagram

`a2a-server.md` — Host A2A Server

Nguồn: how-to/enable-agent-to-agent-endpoint.md Key concepts:

Enable A2A endpoint trên Foundry
Agent Card configuration
Authentication cho A2A

`a2a-connect.md` — Connect to A2A Agent

Nguồn: how-to/enable-agent-to-agent-endpoint.md Key concepts:

Discovery via Agent Card URL
A2A client patterns
Cross-organization agent communication

🔀 Category 05: Orchestration (Module 6, 7, 8)

`workflows-overview.md` — Workflow trong Microsoft Foundry

Nguồn: concepts/workflow.md Key concepts:

Workflow là gì? (orchestration layer bên trên agents)
Workflow vs Agent: phân biệt
Components: Triggers, Actions, Agents, Conditions
Analogy: Workflow như "bản nhạc" — Agent như "nhạc cụ"
Mermaid: Workflow component diagram

`workflow-patterns.md` — Workflow Patterns

Nguồn: concepts/workflow.md Key concepts:

Sequential, Parallel, Event-driven patterns
Khi nào dùng pattern nào?
Mermaid: 3 pattern comparison diagrams

`workflow-foundry.md` — Tạo Workflows trong Foundry

Nguồn: concepts/workflow.md, how-to/vs-code-agents-workflow-low-code.md Key concepts:

Visual workflow builder trong Foundry portal
Low-code approach
Add agents vào workflow
Mermaid: Build workflow step-by-step

`workflow-power-fx.md` — Power Fx trong Workflows

Nguồn: concepts/workflow.md Key concepts:

Power Fx là gì? (Excel-like formula language)
Dùng Power Fx cho conditions & data transformation
Analogy: Power Fx như công thức Excel nhưng cho workflow logic

`agent-framework-intro.md` — Microsoft Agent Framework

Nguồn: concepts/hosted-agents.md Key concepts:

Microsoft Agent Framework (MAF) là gì?
MAF vs LangGraph vs OpenAI Agents SDK: so sánh
Khi nào chọn MAF?
Mermaid: Framework comparison

`agent-framework-create.md` — Tạo Agent với MAF

Nguồn: concepts/hosted-agents.md, quickstarts/quickstart-hosted-agent.md Key concepts:

Agent structure trong MAF
HostedAgent pattern
Responses API integration
Mermaid: MAF agent architecture

`agent-framework-tools.md` — Thêm Tools vào MAF Agent

Nguồn: concepts/tool-best-practice.md, how-to/tools/ Key concepts:

Tool integration patterns trong MAF
Tool selection best practices
Testing tools

`multi-agent-overview.md` — Giới thiệu Multi-Agent Orchestration

Nguồn: concepts/hosted-agents.md Key concepts:

Tại sao Multi-Agent? (specialization, scalability)
Orchestration patterns: Concurrent, Sequential, Group Chat, Handoff, Magentic
Mermaid: Pattern taxonomy diagram

`orchestration-concurrent.md` — Concurrent Orchestration

Key concepts:

Agents chạy song song
Fan-out / Fan-in pattern
Khi nào dùng: tasks độc lập, cần tốc độ

`orchestration-sequential.md` — Sequential Orchestration

Key concepts:

Agents chạy tuần tự, output → input
Pipeline pattern
Khi nào dùng: tasks phụ thuộc nhau

`orchestration-group.md` — Group Chat Orchestration

Key concepts:

Multiple agents trong 1 conversation
Moderator agent pattern
Khi nào dùng: debate, review, brainstorm

`orchestration-handoff.md` — Handoff Orchestration

Key concepts:

Agents "bàn giao" task cho nhau
Triage pattern
Khi nào dùng: routing theo expertise

`orchestration-magentic.md` — Magentic Orchestration

Key concepts:

Magentic = Microsoft's meta-agent pattern
Dynamic planning và task decomposition
Khi nào dùng: phức tạp nhất, tasks không xác định trước

Sidebars.js Configuration

// sidebars.js
const sidebars = {
  azureAiAgentSidebar: [
    {
      type: 'doc',
      id: 'azure-ai-agent/index',
      label: '🤖 Azure AI Agent — Lộ trình',
    },
    {
      type: 'category',
      label: '🏗️ 01. Nền Tảng',
      items: [
        'azure-ai-agent/01-foundations/index',
        'azure-ai-agent/01-foundations/what-is-ai-agent',
        'azure-ai-agent/01-foundations/microsoft-foundry',
        'azure-ai-agent/01-foundations/development-approaches',
        'azure-ai-agent/01-foundations/first-agent-foundry',
        'azure-ai-agent/01-foundations/vscode-setup',
        'azure-ai-agent/01-foundations/agent-tools-overview',
        'azure-ai-agent/01-foundations/deploy-integrate',
      ],
    },
    {
      type: 'category',
      label: '🔧 02. Công Cụ',
      items: [
        'azure-ai-agent/02-tools/index',
        'azure-ai-agent/02-tools/custom-tools-why',
        'azure-ai-agent/02-tools/custom-tools-options',
        'azure-ai-agent/02-tools/custom-tools-integrate',
        'azure-ai-agent/02-tools/mcp-discovery',
        'azure-ai-agent/02-tools/mcp-server-client',
        'azure-ai-agent/02-tools/mcp-azure-agents',
      ],
    },
    {
      type: 'category',
      label: '📚 03. Knowledge & RAG',
      items: [
        'azure-ai-agent/03-knowledge/index',
        'azure-ai-agent/03-knowledge/rag-for-agents',
        'azure-ai-agent/03-knowledge/foundry-iq-overview',
        'azure-ai-agent/03-knowledge/data-sources-config',
        'azure-ai-agent/03-knowledge/retrieval-config',
      ],
    },
    {
      type: 'category',
      label: '🔗 04. Tích Hợp',
      items: [
        'azure-ai-agent/04-integration/index',
        'azure-ai-agent/04-integration/m365-publish-options',
        'azure-ai-agent/04-integration/m365-teams-publish',
        'azure-ai-agent/04-integration/m365-agents-toolkit',
        'azure-ai-agent/04-integration/work-iq',
        'azure-ai-agent/04-integration/a2a-protocol',
        'azure-ai-agent/04-integration/a2a-agent-executor',
        'azure-ai-agent/04-integration/a2a-server',
        'azure-ai-agent/04-integration/a2a-connect',
      ],
    },
    {
      type: 'category',
      label: '🔀 05. Workflow & Orchestration',
      items: [
        'azure-ai-agent/05-orchestration/index',
        'azure-ai-agent/05-orchestration/workflows-overview',
        'azure-ai-agent/05-orchestration/workflow-patterns',
        'azure-ai-agent/05-orchestration/workflow-foundry',
        'azure-ai-agent/05-orchestration/workflow-power-fx',
        'azure-ai-agent/05-orchestration/agent-framework-intro',
        'azure-ai-agent/05-orchestration/agent-framework-create',
        'azure-ai-agent/05-orchestration/agent-framework-tools',
        'azure-ai-agent/05-orchestration/multi-agent-overview',
        'azure-ai-agent/05-orchestration/orchestration-concurrent',
        'azure-ai-agent/05-orchestration/orchestration-sequential',
        'azure-ai-agent/05-orchestration/orchestration-group',
        'azure-ai-agent/05-orchestration/orchestration-handoff',
        'azure-ai-agent/05-orchestration/orchestration-magentic',
      ],
    },
  ],
};

Kế hoạch triển khai (Giai đoạn sau)

Thứ tự ưu tiên viết bài:

Phase 1 — Foundation (Priority 1):

index.md — Landing page
01-foundations/what-is-ai-agent.md ← Bài nền tảng nhất
01-foundations/microsoft-foundry.md
01-foundations/development-approaches.md

Phase 2 — Tools (Priority 2): 5. 02-tools/custom-tools-why.md 6. 02-tools/mcp-discovery.md 7. 02-tools/mcp-server-client.md

Phase 3 — Knowledge (Priority 3): 8. 03-knowledge/rag-for-agents.md 9. 03-knowledge/foundry-iq-overview.md

Phase 4 — Orchestration (Priority 4): 10. 05-orchestration/workflows-overview.md 11. 05-orchestration/multi-agent-overview.md 12. Các bài orchestration patterns

Phase 5 — Integration (Priority 5): 13. 04-integration/m365-publish-options.md 14. 04-integration/a2a-protocol.md 15. Các bài còn lại

Skill sử dụng khi viết:

create-tech-lecture — quản lý toàn bộ template (4MAT, Glossary, Mermaid, Trade-off, TL;DR, Footer). Đây là skill chính cho mỗi bài, không cần định nghĩa template riêng trong plan.
mermaid-expert — tạo diagrams phức tạp, tránh syntax errors
fact-check — verify thông tin từ tài liệu gốc Microsoft
review-report — QA nội dung & format trước khi publish
/push-content workflow — deploy lên GitHub Pages

Tổng quan số lượng

Hạng mục	Số lượng
Categories	5
Files docs	~35
Bài học chính	~30
Diagrams Mermaid ước tính	~90 (3/bài)
Từ tổng ước tính	75.000–120.000 từ
Thời gian đọc tổng	~25–40 giờ

Made by Anh Tu - Share to be share

Tóm tắt dự án​

Quyết định thiết kế​

Cấu trúc thư mục Docs​

Outline chi tiết & Mapping nguồn tài liệu​

🏗️ Category 01: Foundations (Module 1)​

index.md — Giới thiệu Category​

what-is-ai-agent.md — AI Agent là gì?​

microsoft-foundry.md — Microsoft Foundry Agent Service là gì?​

development-approaches.md — Các hướng phát triển Agent​

first-agent-foundry.md — Build Agent đầu tiên trong Foundry Portal​

vscode-setup.md — Setup VS Code cho Agent Development​

agent-tools-overview.md — Mở rộng Agent với Tools​

deploy-integrate.md — Test, Deploy và Integrate Agent​

🔧 Category 02: Tools (Module 2 & 3)​

✅ custom-tools-why.md — Tại sao cần Custom Tools?​

✅ custom-tools-options.md — Các lựa chọn Custom Tools​

✅ custom-tools-integrate.md — Cách tích hợp Custom Tools​

✅ mcp-discovery.md — Understand MCP Tool Discovery​

✅ mcp-server-client.md — MCP Server và Client​

✅ mcp-azure-agents.md — Dùng MCP với Azure AI Agents​

📚 Category 03: Knowledge & RAG (Module 4)​

✅ rag-for-agents.md — RAG cho AI Agents​

✅ foundry-iq-overview.md — Foundry IQ là gì?​

✅ data-sources-config.md — Cấu hình Data Sources​

✅ retrieval-config.md — Cấu hình Retrieval với Foundry IQ​

🔗 Category 04: Integration (Module 5 & 9)​

m365-publish-options.md — Publishing Options trong Foundry​

m365-teams-publish.md — Publish Agent lên Microsoft Teams​

m365-agents-toolkit.md — Microsoft 365 Agents Toolkit (Advanced)​

work-iq.md — Access M365 Data với Work IQ​

a2a-protocol.md — A2A Protocol là gì?​

a2a-agent-executor.md — Implement Agent Executor​

a2a-server.md — Host A2A Server​

a2a-connect.md — Connect to A2A Agent​

🔀 Category 05: Orchestration (Module 6, 7, 8)​

workflows-overview.md — Workflow trong Microsoft Foundry​

workflow-patterns.md — Workflow Patterns​

workflow-foundry.md — Tạo Workflows trong Foundry​

workflow-power-fx.md — Power Fx trong Workflows​

agent-framework-intro.md — Microsoft Agent Framework​

agent-framework-create.md — Tạo Agent với MAF​

agent-framework-tools.md — Thêm Tools vào MAF Agent​

multi-agent-overview.md — Giới thiệu Multi-Agent Orchestration​

orchestration-concurrent.md — Concurrent Orchestration​

orchestration-sequential.md — Sequential Orchestration​

orchestration-group.md — Group Chat Orchestration​

orchestration-handoff.md — Handoff Orchestration​

orchestration-magentic.md — Magentic Orchestration​

Sidebars.js Configuration​

Kế hoạch triển khai (Giai đoạn sau)​

Thứ tự ưu tiên viết bài:​

Skill sử dụng khi viết:​

Tổng quan số lượng​

Tóm tắt dự án

Quyết định thiết kế

Cấu trúc thư mục Docs

Outline chi tiết & Mapping nguồn tài liệu

🏗️ Category 01: Foundations (Module 1)

`index.md` — Giới thiệu Category

`what-is-ai-agent.md` — AI Agent là gì?

`microsoft-foundry.md` — Microsoft Foundry Agent Service là gì?

`development-approaches.md` — Các hướng phát triển Agent

`first-agent-foundry.md` — Build Agent đầu tiên trong Foundry Portal

`vscode-setup.md` — Setup VS Code cho Agent Development

`agent-tools-overview.md` — Mở rộng Agent với Tools

`deploy-integrate.md` — Test, Deploy và Integrate Agent

🔧 Category 02: Tools (Module 2 & 3)

✅ `custom-tools-why.md` — Tại sao cần Custom Tools?

✅ `custom-tools-options.md` — Các lựa chọn Custom Tools

✅ `custom-tools-integrate.md` — Cách tích hợp Custom Tools

✅ `mcp-discovery.md` — Understand MCP Tool Discovery

✅ `mcp-server-client.md` — MCP Server và Client

✅ `mcp-azure-agents.md` — Dùng MCP với Azure AI Agents

📚 Category 03: Knowledge & RAG (Module 4)

✅ `rag-for-agents.md` — RAG cho AI Agents

✅ `foundry-iq-overview.md` — Foundry IQ là gì?

✅ `data-sources-config.md` — Cấu hình Data Sources

✅ `retrieval-config.md` — Cấu hình Retrieval với Foundry IQ

🔗 Category 04: Integration (Module 5 & 9)

`m365-publish-options.md` — Publishing Options trong Foundry

`m365-teams-publish.md` — Publish Agent lên Microsoft Teams

`m365-agents-toolkit.md` — Microsoft 365 Agents Toolkit (Advanced)

`work-iq.md` — Access M365 Data với Work IQ

`a2a-protocol.md` — A2A Protocol là gì?

`a2a-agent-executor.md` — Implement Agent Executor

`a2a-server.md` — Host A2A Server

`a2a-connect.md` — Connect to A2A Agent

🔀 Category 05: Orchestration (Module 6, 7, 8)

`workflows-overview.md` — Workflow trong Microsoft Foundry

`workflow-patterns.md` — Workflow Patterns

`workflow-foundry.md` — Tạo Workflows trong Foundry

`workflow-power-fx.md` — Power Fx trong Workflows

`agent-framework-intro.md` — Microsoft Agent Framework

`agent-framework-create.md` — Tạo Agent với MAF

`agent-framework-tools.md` — Thêm Tools vào MAF Agent

`multi-agent-overview.md` — Giới thiệu Multi-Agent Orchestration

`orchestration-concurrent.md` — Concurrent Orchestration

`orchestration-sequential.md` — Sequential Orchestration

`orchestration-group.md` — Group Chat Orchestration

`orchestration-handoff.md` — Handoff Orchestration

`orchestration-magentic.md` — Magentic Orchestration

Sidebars.js Configuration

Kế hoạch triển khai (Giai đoạn sau)

Thứ tự ưu tiên viết bài:

Skill sử dụng khi viết:

Tổng quan số lượng