faion-ai-agents

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Entry point:
/faion-net
— invoke this skill for automatic routing to the appropriate domain.

入口点：
/faion-net
— 调用此Skill以自动路由到对应领域。

AI Agents Skill

Communication: User's language. Code: English.

沟通语言：用户使用的语言。代码语言：英文。

Purpose

用途

Specializes in AI agent development and orchestration. Covers autonomous agents, multi-agent systems, frameworks, and MCP.

专注于AI Agent的开发与编排。涵盖自主Agent、多Agent系统、相关框架以及MCP。

Context Discovery

上下文发现

Auto-Investigation

自动调查

Check these project signals before asking questions:

Signal	Where to Check	What to Look For
Dependencies	package.json, requirements.txt	langchain, llamaindex, anthropic (MCP)
Agent code	Grep for "agent", "tool", "ReAct"	Existing agent implementations
MCP config	mcp.json, claude_desktop_config.json	MCP servers configuration
Tools/functions	Grep for "function", "tool_def"	Available agent tools

在提问前检查以下项目信号：

信号类型	检查位置	检查内容
依赖项	package.json、requirements.txt	langchain、llamaindex、anthropic（MCP）
Agent代码	搜索"agent"、"tool"、"ReAct"关键词	已有的Agent实现
MCP配置	mcp.json、claude_desktop_config.json	MCP服务器配置
工具/函数	搜索"function"、"tool_def"关键词	可用的Agent工具

Discovery Questions

发现问题

yaml

question: "What type of agent are you building?"
header: "Agent Architecture"
multiSelect: false
options:
  - label: "Single autonomous agent"
    description: "One agent with tools (ReAct, plan-and-execute)"
  - label: "Multi-agent system"
    description: "Multiple agents collaborating/delegating"
  - label: "Agentic RAG"
    description: "Agent-driven document retrieval"
  - label: "MCP integration (Claude tools)"
    description: "Model Context Protocol for Claude Code"

yaml

question: "Which agent framework?"
header: "Framework"
multiSelect: false
options:
  - label: "LangChain"
    description: "Most mature, extensive tooling"
  - label: "LlamaIndex"
    description: "Best for data/document agents"
  - label: "Custom implementation"
    description: "Direct API calls to LLM"
  - label: "Claude MCP (native)"
    description: "Claude-native tool protocol"

yaml

question: "What tools/capabilities does the agent need?"
header: "Agent Capabilities"
multiSelect: true
options:
  - label: "Web search"
    description: "Search internet for information"
  - label: "Code execution"
    description: "Run Python/JS code safely"
  - label: "Database queries"
    description: "Query SQL/NoSQL databases"
  - label: "API calls"
    description: "Call external REST/GraphQL APIs"
  - label: "File operations"
    description: "Read/write files, search codebase"

yaml

question: "What type of agent are you building?"
header: "Agent Architecture"
multiSelect: false
options:
  - label: "Single autonomous agent"
    description: "One agent with tools (ReAct, plan-and-execute)"
  - label: "Multi-agent system"
    description: "Multiple agents collaborating/delegating"
  - label: "Agentic RAG"
    description: "Agent-driven document retrieval"
  - label: "MCP integration (Claude tools)"
    description: "Model Context Protocol for Claude Code"

yaml

question: "Which agent framework?"
header: "Framework"
multiSelect: false
options:
  - label: "LangChain"
    description: "Most mature, extensive tooling"
  - label: "LlamaIndex"
    description: "Best for data/document agents"
  - label: "Custom implementation"
    description: "Direct API calls to LLM"
  - label: "Claude MCP (native)"
    description: "Claude-native tool protocol"

yaml

question: "What tools/capabilities does the agent need?"
header: "Agent Capabilities"
multiSelect: true
options:
  - label: "Web search"
    description: "Search internet for information"
  - label: "Code execution"
    description: "Run Python/JS code safely"
  - label: "Database queries"
    description: "Query SQL/NoSQL databases"
  - label: "API calls"
    description: "Call external REST/GraphQL APIs"
  - label: "File operations"
    description: "Read/write files, search codebase"

Scope

适用范围

Area	Coverage
Agent Patterns	ReAct, plan-and-execute, reasoning-first
Autonomous Agents	Agent loops, memory, tool use
Multi-Agent	Coordination, communication, delegation
Frameworks	LangChain, LlamaIndex agent implementations
MCP	Model Context Protocol, Claude tools
Governance	EU AI Act compliance, safety

领域	覆盖内容
Agent模式	ReAct、计划执行、推理优先
自主Agent	Agent循环、记忆、工具使用
多Agent	协调、通信、任务委派
框架	LangChain、LlamaIndex的Agent实现
MCP	Model Context Protocol、Claude工具
治理	欧盟AI法案合规性、安全性

Quick Start

快速开始

Task	Files
Basic agent	ai-agent-patterns.md → agent-patterns.md
Autonomous agent	autonomous-agents.md → agent-architectures.md
Multi-agent	multi-agent-basics.md → multi-agent-patterns.md
LangChain agents	langchain-agents-architectures.md
MCP integration	mcp-model-context-protocol.md → mcp-ecosystem-2026.md

任务	文件
基础Agent	ai-agent-patterns.md → agent-patterns.md
自主Agent	autonomous-agents.md → agent-architectures.md
多Agent	multi-agent-basics.md → multi-agent-patterns.md
LangChain Agent	langchain-agents-architectures.md
MCP集成	mcp-model-context-protocol.md → mcp-ecosystem-2026.md

Methodologies (26)

方法论（共26种）

Agent Fundamentals (4):

ai-agent-patterns: Core patterns, memory, planning
agent-patterns: ReAct, chain-of-thought, reflection
agent-architectures: System design, components
autonomous-agents: Loops, decision-making, persistence

Multi-Agent (4):

multi-agent-basics: Fundamentals, communication
multi-agent-patterns: Delegation, collaboration
multi-agent-design-patterns: Hierarchical, peer-to-peer

LangChain (7):

langchain-basics: Setup, chains, components
langchain-chains: LCEL, sequential, routing
langchain-memory: Conversation, summary, entity
langchain-workflows: Complex flows, branching
langchain-agents-architectures: Agent types, tools
langchain-agents-multi-agent: Multi-agent with LangChain
langchain-patterns: Production patterns

LlamaIndex (3):

llamaindex-basics: Data connectors, indexes
llamaindex-indexes-queries: Query engines, retrievers
llamaindex-agents-eval: Agent implementation, evaluation

MCP & Tooling (4):

mcp-model-context-protocol: Protocol fundamentals
model-context-protocol: Specification
mcp-ecosystem: Available servers, tools
mcp-ecosystem-2026: Latest developments

Governance (2):

ai-governance-compliance: Frameworks, best practices
eu-ai-act-compliance: Risk tiers, requirements
eu-ai-act-compliance-2026: Latest updates

Advanced (2):

agentic-rag: Agent-driven retrieval (duplicated in RAG)
reasoning-first-architectures: Extended thinking patterns

Agent基础（4种）：

ai-agent-patterns：核心模式、记忆、规划
agent-patterns：ReAct、思维链、反思
agent-architectures：系统设计、组件
autonomous-agents：循环、决策、持久化

多Agent（4种）：

multi-agent-basics：基础原理、通信
multi-agent-patterns：任务委派、协作
multi-agent-design-patterns：分层架构、对等架构

LangChain（7种）：

langchain-basics：环境搭建、链、组件
langchain-chains：LCEL、顺序链、路由链
langchain-memory：对话记忆、摘要记忆、实体记忆
langchain-workflows：复杂流程、分支流程
langchain-agents-architectures：Agent类型、工具
langchain-agents-multi-agent：基于LangChain的多Agent系统
langchain-patterns：生产级模式

LlamaIndex（3种）：

llamaindex-basics：数据连接器、索引
llamaindex-indexes-queries：查询引擎、检索器
llamaindex-agents-eval：Agent实现、评估

MCP与工具（4种）：

mcp-model-context-protocol：协议基础
model-context-protocol：协议规范
mcp-ecosystem：可用服务器、工具
mcp-ecosystem-2026：最新发展

治理（2种）：

ai-governance-compliance：框架、最佳实践
eu-ai-act-compliance：风险等级、要求
eu-ai-act-compliance-2026：最新更新

进阶内容（2种）：

agentic-rag：Agent驱动的检索（与RAG技能重复）
reasoning-first-architectures：扩展思维模式

Agent Architectures

Agent架构

ReAct Pattern

ReAct模式

Input → Thought → Action → Observation → Thought → ... → Answer

Input → Thought → Action → Observation → Thought → ... → Answer

Plan-and-Execute

计划执行模式

Input → Plan → Execute Step 1 → Execute Step 2 → ... → Synthesize

Input → Plan → Execute Step 1 → Execute Step 2 → ... → Synthesize

Reasoning-First

推理优先模式

Input → Extended Thinking → Plan → Execute → Answer

Input → Extended Thinking → Plan → Execute → Answer

Code Examples

代码示例

Basic ReAct Agent (LangChain)

基础ReAct Agent（基于LangChain）

python

from langchain.agents import create_react_agent, AgentExecutor
from langchain_openai import ChatOpenAI
from langchain.tools import Tool

tools = [
    Tool(
        name="Calculator",
        func=lambda x: eval(x),
        description="Math calculator"
    )
]

llm = ChatOpenAI(model="gpt-4o")
agent = create_react_agent(llm, tools, prompt)
executor = AgentExecutor(agent=agent, tools=tools)

result = executor.invoke({"input": "What is 25 * 17?"})

python

from langchain.agents import create_react_agent, AgentExecutor
from langchain_openai import ChatOpenAI
from langchain.tools import Tool

tools = [
    Tool(
        name="Calculator",
        func=lambda x: eval(x),
        description="Math calculator"
    )
]

llm = ChatOpenAI(model="gpt-4o")
agent = create_react_agent(llm, tools, prompt)
executor = AgentExecutor(agent=agent, tools=tools)

result = executor.invoke({"input": "What is 25 * 17?"})

Multi-Agent System

多Agent系统

python

from langchain.agents import initialize_agent, Tool
from langchain_openai import ChatOpenAI

python

from langchain.agents import initialize_agent, Tool
from langchain_openai import ChatOpenAI

Define specialized agents

researcher = ChatOpenAI(model="gpt-4o") writer = ChatOpenAI(model="gpt-4o")

Orchestrator delegates tasks

orchestrator = initialize_agent( tools=[ Tool(name="research", func=research_agent), Tool(name="write", func=writer_agent) ], llm=ChatOpenAI(model="gpt-4o"), agent="zero-shot-react-description" )

result = orchestrator.invoke("Research AI trends and write a summary")

undefined

orchestrator = initialize_agent( tools=[ Tool(name="research", func=research_agent), Tool(name="write", func=writer_agent) ], llm=ChatOpenAI(model="gpt-4o"), agent="zero-shot-react-description" )

result = orchestrator.invoke("Research AI trends and write a summary")

undefined

MCP Server Integration

MCP服务器集成

python

import anthropic

client = anthropic.Anthropic()

response = client.messages.create(
    model="claude-sonnet-4-20250514",
    max_tokens=1024,
    tools=[{
        "name": "get_weather",
        "description": "Get weather data",
        "input_schema": {
            "type": "object",
            "properties": {
                "location": {"type": "string"}
            }
        }
    }],
    messages=[{"role": "user", "content": "Weather in NYC?"}]
)

python

import anthropic

client = anthropic.Anthropic()

response = client.messages.create(
    model="claude-sonnet-4-20250514",
    max_tokens=1024,
    tools=[{
        "name": "get_weather",
        "description": "Get weather data",
        "input_schema": {
            "type": "object",
            "properties": {
                "location": {"type": "string"}
            }
        }
    }],
    messages=[{"role": "user", "content": "Weather in NYC?"}]
)

LlamaIndex Agent

python

from llama_index.agent import ReActAgent
from llama_index.llms import OpenAI
from llama_index.tools import QueryEngineTool

llm = OpenAI(model="gpt-4o")

tools = [
    QueryEngineTool.from_defaults(
        query_engine=query_engine,
        name="docs",
        description="Documentation search"
    )
]

agent = ReActAgent.from_tools(tools, llm=llm)
response = agent.chat("How do I use embeddings?")

python

from llama_index.agent import ReActAgent
from llama_index.llms import OpenAI
from llama_index.tools import QueryEngineTool

llm = OpenAI(model="gpt-4o")

tools = [
    QueryEngineTool.from_defaults(
        query_engine=query_engine,
        name="docs",
        description="Documentation search"
    )
]

agent = ReActAgent.from_tools(tools, llm=llm)
response = agent.chat("How do I use embeddings?")

Multi-Agent Patterns

多Agent模式

Pattern	Use Case
Hierarchical	Manager delegates to specialists
Peer-to-Peer	Agents collaborate as equals
Sequential	Chain of agents, each refines
Parallel	Multiple agents work simultaneously

模式	适用场景
分层模式	管理者将任务委派给专业Agent
对等模式	Agent之间平等协作
顺序模式	多Agent链式处理，逐步优化结果
并行模式	多Agent同时处理任务

MCP Ecosystem (2026)

MCP生态系统（2026版）

Server	Purpose
filesystem	File operations
postgres	Database queries
puppeteer	Web automation
github	GitHub API access
slack	Slack integration

服务器	用途
filesystem	文件操作
postgres	数据库查询
puppeteer	网页自动化
github	GitHub API访问
slack	Slack集成

EU AI Act Compliance

欧盟AI法案合规性

Risk Tier	Requirements
Unacceptable	Banned (social scoring, manipulation)
High-risk	Conformity assessment, documentation
Limited-risk	Transparency obligations
Minimal-risk	No obligations

风险等级	要求
不可接受风险	禁止使用（如社会评分、操纵类应用）
高风险	合规性评估、文档记录
有限风险	透明度要求
低风险	无强制要求

Skill	Relationship
faion-llm-integration	Provides LLM APIs
faion-rag-engineer	Agentic RAG integration
faion-ml-ops	Agent evaluation

Skill	关联关系
faion-llm-integration	提供LLM API
faion-rag-engineer	Agentic RAG集成
faion-ml-ops	Agent评估

faion-ai-agents

Original

Translation

AI Agents Skill

AI Agents Skill

Purpose

用途

Context Discovery

上下文发现

Auto-Investigation

自动调查

Discovery Questions

发现问题

Scope

适用范围

Quick Start

快速开始

Methodologies (26)

方法论（共26种）

Agent Architectures

Agent架构

ReAct Pattern

ReAct模式

Plan-and-Execute

计划执行模式

Reasoning-First

推理优先模式

Code Examples

代码示例

Basic ReAct Agent (LangChain)

基础ReAct Agent（基于LangChain）

Multi-Agent System

多Agent系统

Define specialized agents

Define specialized agents

Orchestrator delegates tasks

Orchestrator delegates tasks

MCP Server Integration

MCP服务器集成

LlamaIndex Agent

LlamaIndex Agent

Multi-Agent Patterns

多Agent模式

MCP Ecosystem (2026)

MCP生态系统（2026版）

EU AI Act Compliance

欧盟AI法案合规性

Related Skills

相关Skill