10分钟上手

1. 启动服务

python -m venv .venv
source .venv/bin/activate
pip install -e '.[dev,docs]'
uvicorn industrial_agent_flow.api.app:app --reload --port 8000

2. 配置模型

export INDUSTRIAL_AGENT_FLOW_LLM_VENDOR=qwen
export INDUSTRIAL_AGENT_FLOW_LLM_API_KEY=your_api_key
export INDUSTRIAL_AGENT_FLOW_LLM_MODEL_PARSER=qwen-plus
export INDUSTRIAL_AGENT_FLOW_LLM_MODEL_GENERATOR=qwen-plus
export INDUSTRIAL_AGENT_FLOW_LLM_MODEL_CRITIC=qwen-plus
export INDUSTRIAL_AGENT_FLOW_LLM_FAILURE_MODE=auto

3. 导入 YAML

curl -X POST http://127.0.0.1:8000/v1/catalogs/import \
  -H 'content-type: application/json' \
  -d '{"mode":"legacy","yaml_path":"/Users/shunyu/Downloads/ClosedLoopOptimization/mechanism_v1.yml"}'

返回结果中会有 catalog_id，后续直接使用它。

4. 基于 catalog 运行迁移

curl -X POST http://127.0.0.1:8000/v1/agentic/run-from-catalog \
  -H 'content-type: application/json' \
  -d '{
    "catalog_id":"legacy-scene-catalog",
    "nl_requirements":["在安全边界内最小化能耗","主蒸汽压力不低于30"],
    "max_iterations":3
  }'

5. 校验和发布

curl -X POST http://127.0.0.1:8000/v1/templates/validate \
  -H 'content-type: application/json' \
  -d @draft.json

curl -X POST http://127.0.0.1:8000/v1/templates/quality-check \
  -H 'content-type: application/json' \
  -d '{"draft": <draft_object>}'

curl -X POST http://127.0.0.1:8000/v1/templates/publish \
  -H 'content-type: application/json' \
  -d '{"draft": <draft_object>}'

6. 仿真与评测

curl -X POST http://127.0.0.1:8000/v1/pipeline/simulate \
  -H 'content-type: application/json' \
  -d '{"template_id":"line-a-template","scene_context":{"values":{"energy_cost":120.0},"metadata":{}}}'