# 克隆代码
git clone https://github.com/catface996/hierarchical-agents.git
cd hierarchical-agents

# 配置环境变量
cp .env.example .env
# 编辑 .env 文件，配置 AWS 认证

# 启动服务
docker-compose up -d

# 验证服务
curl http://localhost:8080/health

# 拉取镜像 (自动选择正确架构)
docker pull catface996/hierarchical-agents:latest

# 使用生产配置启动
docker-compose -f docker-compose.prod.yml up -d

# 安装依赖
pip install -r requirements.txt

# 配置环境变量
export AWS_BEDROCK_API_KEY='your-api-key'
export AWS_REGION='us-east-1'

# 启动服务
python -m src.ec2.server

# API Key 认证
export AWS_BEDROCK_API_KEY='your-api-key'

# 或 AK/SK 认证
export AWS_ACCESS_KEY_ID='your-ak'
export AWS_SECRET_ACCESS_KEY='your-sk'

# 或 IAM Role 认证 (EC2/Lambda)
export USE_IAM_ROLE=true

curl -X POST http://localhost:8080/api/v1/hierarchies/create \
  -H "Content-Type: application/json" \
  -d '{
    "name": "研究团队",
    "global_prompt": "你是研究团队的协调者",
    "teams": [
      {
        "name": "分析组",
        "supervisor_prompt": "你是分析组负责人",
        "workers": [
          {"name": "分析师", "role": "数据分析", "system_prompt": "你是数据分析专家"}
        ]
      }
    ]
  }'

curl -X POST http://localhost:8080/api/v1/runs/start \
  -H "Content-Type: application/json" \
  -d '{
    "hierarchy_id": "your-hierarchy-id",
    "task": "分析这个问题..."
  }'

# 自动获取第一个层级团队
python test_stream.py "你的问题"

# 指定层级团队
python test_stream.py --hierarchy=ID "你的问题"

# 指定远程 API
python test_stream.py --api=http://ec2-ip:8080 "你的问题"

# Step 0: 获取可用模型列表，选择一个模型 ID
curl -s -X POST http://localhost:8080/api/v1/models/list \
  -H "Content-Type: application/json" -d '{}' | jq '.data.items[] | {id, name}'

# 示例输出:
# { "id": "abc123-...", "name": "Claude Sonnet 4" }
# { "id": "def456-...", "name": "Claude Opus 4" }

# 设置模型 ID (替换为实际的模型 ID)
MODEL_ID="your-model-id-here"

# Step 1: 创建层级团队 (仅需一次)
HIERARCHY_ID=$(curl -s -X POST http://localhost:8080/api/v1/hierarchies/create \
  -H "Content-Type: application/json" \
  -d "{
    \"name\": \"研究团队\",
    \"global_prompt\": \"你是研究协调者，负责分析任务并分配给合适的团队\",
    \"llm_config\": {\"model_id\": \"$MODEL_ID\", \"temperature\": 0.7, \"max_tokens\": 4096},
    \"teams\": [{
      \"name\": \"分析组\",
      \"supervisor_prompt\": \"你是分析组负责人\",
      \"llm_config\": {\"model_id\": \"$MODEL_ID\", \"temperature\": 0.6},
      \"workers\": [{
        \"name\": \"分析师\",
        \"role\": \"数据分析专家\",
        \"system_prompt\": \"你是数据分析专家，擅长深度分析问题\",
        \"llm_config\": {\"model_id\": \"$MODEL_ID\", \"temperature\": 0.5}
      }]
    }]
  }" | jq -r '.data.id')

echo "Created hierarchy: $HIERARCHY_ID"

# Step 2: 启动任务执行
RUN_ID=$(curl -s -X POST http://localhost:8080/api/v1/runs/start \
  -H "Content-Type: application/json" \
  -d "{\"hierarchy_id\": \"$HIERARCHY_ID\", \"task\": \"分析人工智能的发展趋势\"}" \
  | jq -r '.data.id')

echo "Started run: $RUN_ID"

# Step 3: 流式获取执行过程 (SSE) - 需要在启动后立即调用
curl -N -X POST http://localhost:8080/api/v1/runs/stream \
  -H "Content-Type: application/json" \
  -d "{\"id\": \"$RUN_ID\"}"

# Step 4: 或者轮询获取结果
curl -s -X POST http://localhost:8080/api/v1/runs/get \
  -H "Content-Type: application/json" \
  -d "{\"id\": \"$RUN_ID\"}" | jq

# 登录 Docker Hub
docker login

# 构建并推送多平台镜像
./build-multiarch.sh

# 或手动构建
docker buildx build --platform linux/amd64,linux/arm64 \
  -t catface996/hierarchical-agents:latest \
  --push .