# uv（官方推荐）
uv venv --python 3.12
source .venv/bin/activate          # Windows: .venv\Scripts\Activate.ps1
uv pip install -U agno openai      # 或 uv pip install -U "agno[os]" 装 AgentOS 运行时

# pip 等价
python3 -m venv .venv && source .venv/bin/activate
pip install -U agno openai

mkdir personal_agent && cd personal_agent
uv init --bare --python 3.14
uv add "agno[os,sqlite]" openai
export OPENAI_API_KEY="sk-..."     # Windows: $env:OPENAI_API_KEY="sk-..."

# personal_agent.py
from agno.agent import Agent
from agno.db.sqlite import SqliteDb
from agno.os import AgentOS

db = SqliteDb(db_file="personal_agent.db")

agent = Agent(
    name="Pip",
    model="openai:gpt-5.6",              # 字符串简写，等价 OpenAIResponses(id="gpt-5.6")
    db=db,                                # 会话/记忆/追踪存储
    instructions=["帮用户跟踪项目与任务，回复简短直接"],
    add_history_to_context=True,          # 多轮对话历史注入
    add_datetime_to_context=True,
    enable_agentic_memory=True,           # Agent 自主管理用户记忆
    markdown=True,
)

agent_os = AgentOS(agents=[agent], db=db, tracing=True)
app = agent_os.get_app()

if __name__ == "__main__":
    agent_os.serve(app="personal_agent:app", reload=True)

uv run personal_agent.py
# AgentOS 运行在 http://localhost:7777 ，OpenAPI 在 /docs

from agno.agent import Agent
from agno.models.openai import OpenAIChat, OpenAIResponses

agent = Agent(
    model=OpenAIChat(id="gpt-4.1"),
    reasoning_model=OpenAIResponses(id="gpt-5.2", reasoning_effort="medium"),
)

git clone https://github.com/agno-agi/agentos-railway.git agent-platform
cd agent-platform
cp example.env .env
# 编辑 .env：至少设置 OPENAI_API_KEY（模板默认还需 ANTHROPIC_API_KEY）
docker compose up -d --build
# 两个容器：FastAPI 应用（宿主 8000）+ Postgres/pgvector（127.0.0.1:5432）
# 打开 http://localhost:8000/docs 查看全部 REST 端点
./scripts/mcp_check.sh   # 校验 MCP server（预期输出 9 tools; required tools available）

services:
  agentos-db:
    image: agnohq/pgvector:18
    ports: ["127.0.0.1:5432:5432"]
    environment:
      POSTGRES_USER: ${DB_USER:-ai}
      POSTGRES_PASSWORD: ${DB_PASS:-ai}
      POSTGRES_DB: ${DB_DATABASE:-ai}
  agentos-api:
    build: { context: ., dockerfile: Dockerfile, target: agentos }
    command: uvicorn app.main:app --host 0.0.0.0 --port 8000 --reload
    ports: ["127.0.0.1:8000:8000"]
    environment:
      WAIT_FOR_DB: "True"
      DB_HOST: agentos-db
      AGENTOS_URL: http://127.0.0.1:8000

# 前置：uv 与 Railway CLI 5.49.0+，railway login 已执行
./scripts/venv_setup.sh
[ -f .env.production ] || cp example.env .env.production
# 编辑 .env.production：OPENAI_API_KEY 等
./scripts/railway/up.sh --workspace YOUR_WORKSPACE
# 首次部署会创建 API 服务 + Postgres(pgvector) + 公网域名，然后停下等你配 JWT（退出码 2，属预期）

./scripts/railway/env-sync.sh --no-deploy
./scripts/railway/redeploy.sh
railway logs --service agent-os
# 验证：/health 返回正常；无 token 请求 /agents 被拒绝、带 token 正常

FROM agnohq/python:3.14
WORKDIR /app
COPY pyproject.toml uv.lock ./
RUN uv sync --python 3.14 --locked --no-dev --no-install-project
COPY personal_agent.py ./
CMD ["sh", "-c", "exec uv run --no-sync uvicorn personal_agent:app --host 0.0.0.0 --port ${PORT:-8000}"]

import os
from agno.db.postgres import PostgresDb, create_postgres_engine
from agno.db.sqlite import SqliteDb

if os.getenv("DATABASE_URL"):
    db = PostgresDb(db_engine=create_postgres_engine(os.environ["DATABASE_URL"]))
else:
    db = SqliteDb(db_file="personal_agent.db")

agent_os = AgentOS(
    agents=[agent], db=db,
    authorization=bool(os.getenv("JWT_VERIFICATION_KEY")),  # 有公钥才开 JWT
    tracing=True,
)

git clone https://github.com/agno-agi/agno.git
cd agno
# 各库在 libs/ 下（libs/agno 为主 SDK），用 uv 管理工作区
uv sync
uv run pytest libs/agno/tests   # 测试（以仓库 CONTRIBUTING.md 为准）

from agno.agent import Agent
from agno.db.postgres import PostgresDb
from agno.os import AgentOS

db = PostgresDb(db_url="postgresql+psycopg://ai:ai@localhost:5432/ai")

agent = Agent(
    id="support-agent",
    name="Support Agent",
    model="openai:gpt-5.6",
    db=db,
    enable_agentic_memory=True,
    add_history_to_context=True,
    num_history_runs=3,
    markdown=True,
)

agent_os = AgentOS(
    id="my-os",
    agents=[agent],
    db=db,                       # 追踪/调度/服务账号也落这个库
    tracing=True,                # OpenTelemetry 追踪入库
    authorization=True,          # 读 JWT_VERIFICATION_KEY 或 JWKS
    mcp=True,                    # 挂载 /mcp MCP 服务器
    scheduler=True,              # 启用 cron 调度
    telemetry=False,             # 关闭 AgentOS 遥测
    cors_allowed_origins=["https://app.example.com"],
)
app = agent_os.get_app()

if __name__ == "__main__":
    agent_os.serve(app="main:app", host="0.0.0.0", port=7777)

chat:
  quick_prompts:                    # Chat 页快捷提示（按 agent id）
    support-agent:
      - "我能做什么？"
      - "我的工单进度如何？"
memory:
  display_name: "用户记忆"
  dbs:
    - db_id: db-0001
      domain_config:
        display_name: 主站用户记忆
knowledge:
  display_name: "知识库"
evals:
  display_name: "回归评估"

python personal_agent.py                       # AgentOS.serve() 内部起 uvicorn，默认 7777
# 或直接 uvicorn：
uvicorn personal_agent:app --host 0.0.0.0 --port 7777
export AGENT_OS_HOST=0.0.0.0 AGENT_OS_PORT=8000   # 环境变量优先于 serve() 参数

# /etc/systemd/system/agno-agentos.service
[Unit]
Description=Agno AgentOS Service
After=network.target postgresql.service

[Service]
User=agno
WorkingDirectory=/opt/agno-app
EnvironmentFile=/opt/agno-app/.env
ExecStart=/opt/agno-app/.venv/bin/uvicorn personal_agent:app --host 0.0.0.0 --port 7777 --workers 2
Restart=always
RestartSec=5

[Install]
WantedBy=multi-user.target

sudo systemctl daemon-reload && sudo systemctl enable --now agno-agentos

nssm install AgnoAgentOS "C:\apps\agno\.venv\Scripts\uvicorn.exe" "personal_agent:app --host 0.0.0.0 --port 7777"
nssm set AgnoAgentOS AppDirectory C:\apps\agno
nssm set AgnoAgentOS AppEnvironmentExtra OPENAI_API_KEY=sk-*** AGENT_OS_PORT=7777
nssm start AgnoAgentOS

server {
    listen 443 ssl;
    server_name agents.example.com;
    location / {
        proxy_pass http://127.0.0.1:7777;
        proxy_http_version 1.1;
        proxy_set_header Connection "";        # SSE 流式必需
        proxy_set_header Host $host;
        proxy_read_timeout 300s;               # 长 run 防中断
    }
}

from agno.agent import Agent
from agno.models.anthropic import Claude
from agno.tools.hackernews import HackerNewsTools

agent = Agent(
    model=Claude(id="claude-sonnet-4-5"),
    tools=[HackerNewsTools()],
    instructions="Write a report on the topic. Output only the report.",
    markdown=True,
)
agent.print_response("Trending startups and products.", stream=True)

uv pip install -U agno anthropic
export ANTHROPIC_API_KEY="sk-ant-..."
python hackernews_agent.py

from agno.agent import Agent
from agno.db.sqlite import SqliteDb

agent = Agent(
    model="openai:gpt-5.6",
    db=SqliteDb(db_file="agent.db"),
    add_history_to_context=True,
    num_history_runs=3,
    enable_agentic_memory=True,     # Agent 会把"用户偏好"写进记忆表
)
agent.print_response("I'm working on a Python API project", session_id="dev_session")
agent.print_response("What testing framework should I use?", session_id="dev_session")
# 第二次提问能接住上下文；换 session_id 但同 user_id 时记忆仍在

from agno.team import Team, TeamMode
from agno.agent import Agent
from agno.models.openai import OpenAIResponses
from agno.tools.hackernews import HackerNewsTools
from agno.tools.yfinance import YFinanceTools

news_agent = Agent(id="news-agent", name="News Agent",
                   role="Get trending tech news from HackerNews", tools=[HackerNewsTools()])
finance_agent = Agent(id="finance-agent", name="Finance Agent",
                      role="Get stock prices and financial data", tools=[YFinanceTools()])

team = Team(
    name="Research Team",
    members=[news_agent, finance_agent],
    model=OpenAIResponses(id="gpt-5.4-mini"),
    mode=TeamMode.coordinate,      # 默认：Leader 分解-委派-汇总
    # route：按问题路由给单个专家直接回答；broadcast：全员同题后汇总
    # tasks：任务循环，max_iterations=6 控制轮数
    instructions="Delegate to the appropriate agent based on the request.",
)
team.print_response("What are the trending AI stories and how is NVDA stock doing?", stream=True)

# 1. 起 pgvector 容器（官方镜像）
docker run -d \
  -e POSTGRES_DB=ai -e POSTGRES_USER=ai -e POSTGRES_PASSWORD=ai \
  -e PGDATA=/var/lib/postgresql -v pgvolume:/var/lib/postgresql \
  -p 5532:5432 --name pgvector agnohq/pgvector:18

# 2. 装依赖
uv pip install -U agno openai sqlalchemy psycopg pgvector pypdf

from agno.agent import Agent
from agno.knowledge.embedder.openai import OpenAIEmbedder
from agno.knowledge.knowledge import Knowledge
from agno.models.openai import OpenAIResponses
from agno.vectordb.pgvector import PgVector, SearchType

knowledge = Knowledge(
    vector_db=PgVector(
        table_name="recipes",
        db_url="postgresql+psycopg://ai:ai@localhost:5532/ai",
        search_type=SearchType.hybrid,     # 向量+关键词混合检索
        embedder=OpenAIEmbedder(id="text-embedding-3-small"),
    ),
)
knowledge.insert(url="https://agno-public.s3.amazonaws.com/recipes/ThaiRecipes.pdf")

agent = Agent(
    model=OpenAIResponses(id="gpt-5.2"),
    knowledge=knowledge,
    search_knowledge=True,     # Agentic RAG：模型自主决定何时检索
    markdown=True,
)
agent.print_response("How do I make chicken and galangal in coconut milk soup", stream=True)

# 发现实例（公开端点，返回 auth_mode 与组件数量）
curl http://localhost:7777/info

# 跑一个 Agent（表单字段）
curl http://localhost:7777/agents/support-agent/runs \
  -H "Content-Type: application/x-www-form-urlencoded" \
  -d "message=Where is my order?" \
  -d "user_id=customer-42" \
  -d "session_id=order-support-42" \
  -d "stream=false"

# SSE 流式
curl -N http://localhost:7777/agents/support-agent/runs \
  -d "message=Investigate this account" -d "stream=true"

# 带运行时上下文（JSON 编码的表单字段）
curl http://localhost:7777/agents/story-writer/runs \
  -d "message=Write a short story" \
  -d 'dependencies={"reader_age":10}' \
  -d 'output_schema={"type":"object","properties":{"title":{"type":"string"},"story":{"type":"string"}},"required":["title","story"]}' \
  -d "stream=false"

import asyncio
from agno.client import AgentOSClient

async def main():
    client = AgentOSClient(base_url="http://localhost:7777")
    resp = await client.run_agent(agent_id="support-agent", message="Where is my order?",
                                  user_id="customer-42", session_id="order-support-42")
    print(resp.content)

asyncio.run(main())

# 开源 Agent UI（Next.js）
npx create-agent-ui@latest
cd agent-ui && npm run dev      # 或 git clone https://github.com/agno-agi/agent-ui.git && pnpm install && pnpm dev
# 打开 http://localhost:3000 ，左侧填 AgentOS 端点 http://localhost:7777 即可对话

# 把外部 MCP 服务器接成 Agent 工具（Streamable HTTP）
from agno.tools.mcp import MCPTools

async def run():
    async with MCPTools(url="https://your-mcp-server/mcp", transport="streamable_http") as tools:
        agent = Agent(model="openai:gpt-5.6", tools=[tools])
        await agent.aprint_response("帮我查一下……")

# 方式一：BYO FastAPI —— 把 AgentOS 挂进你已有的应用
from fastapi import FastAPI
from agno.os import AgentOS

app = FastAPI()          # 你现有的路由、中间件都在
agent_os = AgentOS(agents=[agent], base_app=app)   # 路由冲突可用 on_route_conflict="preserve_base_app"

# 方式二：Streamlit 前端调用 AgentOS REST
import streamlit as st
import requests

st.title("我的产品助手")
if prompt := st.chat_input("输入问题"):
    r = requests.post(
        "http://localhost:7777/agents/product-agent/runs",
        data={"message": prompt, "user_id": st.session_state.get("uid", "anon"), "stream": "false"},
    )
    st.markdown(r.json().get("content", ""))

pip show agno                       # SDK 版本
uv pip list | grep agno
curl http://localhost:7777/info     # 返回 agno_version 字段
docker compose images               # 模板镜像版本

pip install -U agno            # 或 uv pip install -U agno
uv lock && uv sync             # 模板项目：更新 uv.lock 后重建
docker compose up -d --build   # Docker 模板升级

pip install agno==3.0.6        # 锁回指定版本（以 PyPI 实际版本为准）
git checkout <旧 tag>          # 模板仓库回滚代码
docker compose up -d --build   # 用旧代码重建
# Railway：redeploy 旧的成功部署，或 git revert 后重推

# 应用
pip uninstall agno -y          # 或 uv pip uninstall agno
rm -rf .venv personal_agent.db tmp/agents.db data/

# Docker 模板
docker compose down -v         # 删除 agentos-api / agentos-db 与 pgdata 卷（毁数据）
docker rmi agnohq/pgvector:18 agnohq/python:3.14 2>/dev/null
docker image prune -f

# Agent UI
rm -rf agent-ui                # Node 项目直接删目录

# Railway
railway open                  # 控制台删除项目（应用+数据库+数据一并删除）
# systemd 服务：sudo systemctl disable --now agno-agentos && sudo rm /etc/systemd/system/agno-agentos.service

# ---- 安装 ----
uv pip install -U agno openai                 # SDK + OpenAI 提供商
uv pip install -U "agno[os]"                  # AgentOS 运行时
uv add "agno[os,sqlite,postgres,pgvector]"    # uv 项目方式
export OPENAI_API_KEY=sk-***

# ---- 本地运行 ----
python personal_agent.py                      # AgentOS 默认 http://localhost:7777
export AGENT_OS_HOST=0.0.0.0 AGENT_OS_PORT=8000
uvicorn personal_agent:app --host 0.0.0.0 --port 7777

# ---- Postgres / pgvector ----
docker run -d -e POSTGRES_DB=ai -e POSTGRES_USER=ai -e POSTGRES_PASSWORD=ai \
  -e PGDATA=/var/lib/postgresql -v pgvolume:/var/lib/postgresql \
  -p 5532:5432 --name pgvector agnohq/pgvector:18

# ---- 平台模板（Docker） ----
git clone https://github.com/agno-agi/agentos-railway.git agent-platform
cd agent-platform && cp example.env .env      # 填 OPENAI_API_KEY
docker compose up -d --build                  # http://localhost:8000/docs
./scripts/mcp_check.sh

# ---- Railway 部署 ----
./scripts/venv_setup.sh
cp example.env .env.production                # 填生产 Key
./scripts/railway/up.sh --workspace YOUR_WORKSPACE
./scripts/railway/env-sync.sh --no-deploy     # 配好 JWT_VERIFICATION_KEY 后
./scripts/railway/redeploy.sh
railway logs --service agent-os

# ---- REST API ----
curl http://localhost:7777/info
curl http://localhost:7777/agents/<agent_id>/runs \
  -d "message=hello" -d "user_id=u1" -d "session_id=s1" -d "stream=false"

# ---- Agent UI ----
npx create-agent-ui@latest && cd agent-ui && npm run dev   # http://localhost:3000

# ---- 升级/回滚/卸载 ----
pip install -U agno
pip install agno==<version>
docker compose down -v && docker image prune -f
pip uninstall agno -y && rm -rf .venv *.db