【AI】04AI Aent:十分钟跑通LangGraph项目:调用llm+agent开发+langSmith使用

Biehmltym

505人浏览 · 2025-12-29 17:41:00

Biehmltym · 2025-12-29 17:41:00 发布

langGraph

目录标题

1.conda python虚拟环境创建
2.安装 langgraph-cli
3.创建langgraph-cli应用
- 总结 CLI 实际完成的事情是：
5.用 PyCharm 打开创建的项目
6.langgraph项目解析
7. pip install -e . 读取 pyproject.toml在虚拟环境安装依赖
8. .env配置环境文件：API KEY安装 llm对应的 LangChain Provider
- test_gemini.py 项目目录下，可以新预先看llm能否跑通![在这里插入图片描述](https://i-blog.csdnimg.cn/direct/3ed5744f7ad8479c8d4a7ca4ec5874bc.png)
9.graph.py：AI Agent智能体的业务代码
10.langgraph dev启动 LangGraph 开发服务器
11.langSmith 使用
12.langSmith 查看agent是否跑通

)

1.conda python虚拟环境创建

conda create --name new_langgrapg_env  python=3.11

实操
先退出现有的langchain虚拟环境
然后新建一个langgraph的虚拟环境3.11


C:\Users\HP>conda activate zyyenv

(zyyenv) C:\Users\HP>python  --version
Python 3.13.11

(zyyenv) C:\Users\HP>conda deactivate

C:\Users\HP>conda create --name new_langgrapg_env  python=3.11
3 channel Terms of Service accepted
Retrieving notices: done
Channels:
 - defaults
Platform: win-64
Collecting package metadata (repodata.json): done
Solving environment: done

## Package Plan ##

  environment location: C:\Users\HP\miniconda3\envs\new_langgrapg_env

  added / updated specs:
    - python=3.11


The following packages will be downloaded:

    package                    |            build
    ---------------------------|-----------------
    expat-2.7.3                |       h885b0b7_4          31 KB
    python-3.11.14             |       h981015d_0        17.7 MB
    setuptools-80.9.0          |  py311haa95532_0         1.9 MB
    wheel-0.45.1               |  py311haa95532_0         182 KB
    ------------------------------------------------------------
                                           Total:        19.8 MB

The following NEW packages will be INSTALLED:

  bzip2              pkgs/main/win-64::bzip2-1.0.8-h2bbff1b_6
  ca-certificates    pkgs/main/win-64::ca-certificates-2025.12.2-haa95532_0
  expat              pkgs/main/win-64::expat-2.7.3-h885b0b7_4
  libexpat           pkgs/main/win-64::libexpat-2.7.3-h885b0b7_4
  libffi             pkgs/main/win-64::libffi-3.4.4-hd77b12b_1
  libzlib            pkgs/main/win-64::libzlib-1.3.1-h02ab6af_0
  openssl            pkgs/main/win-64::openssl-3.0.18-h543e019_0
  pip                pkgs/main/noarch::pip-25.3-pyhc872135_0
  python             pkgs/main/win-64::python-3.11.14-h981015d_0
  setuptools         pkgs/main/win-64::setuptools-80.9.0-py311haa95532_0
  sqlite             pkgs/main/win-64::sqlite-3.51.0-hda9a48d_0
  tk                 pkgs/main/win-64::tk-8.6.15-hf199647_0
  tzdata             pkgs/main/noarch::tzdata-2025b-h04d1e81_0
  ucrt               pkgs/main/win-64::ucrt-10.0.22621.0-haa95532_0
  vc                 pkgs/main/win-64::vc-14.3-h2df5915_10
  vc14_runtime       pkgs/main/win-64::vc14_runtime-14.44.35208-h4927774_10
  vs2015_runtime     pkgs/main/win-64::vs2015_runtime-14.44.35208-ha6b5a95_10
  wheel              pkgs/main/win-64::wheel-0.45.1-py311haa95532_0
  xz                 pkgs/main/win-64::xz-5.6.4-h4754444_1
  zlib               pkgs/main/win-64::zlib-1.3.1-h02ab6af_0


Proceed ([y]/n)? y


Downloading and Extracting Packages:
python-3.11.14       | 17.7 MB   | ############################################################################ | 100%
setuptools-80.9.0    | 1.9 MB    | ############################################################################ | 100%
wheel-0.45.1         | 182 KB    | ############################################################################ | 100%

Preparing transaction: done
Verifying transaction: done
Executing transaction: done
#
# To activate this environment, use
#
#     $ conda activate new_langgrapg_env
#
# To deactivate an active environment, use
#
#     $ conda deactivate


C:\Users\HP>

激活虚拟环境

#
# To activate this environment, use
#
#     $ conda activate new_langgrapg_env
#
# To deactivate an active environment, use
#
#     $ conda deactivate


C:\Users\HP>conda activate new_langgrapg_env

(new_langgrapg_env) C:\Users\HP>

电脑中实际位置

C:\Users\HP\miniconda3\envs\new_langgrapg_env\Scripts

在这里插入图片描述

2.安装 langgraph-cli

官方推荐命令来安装langgraph-cli ，未指定版本即为最新版

pip install --upgrade "langgraph-cli[inmem]"

实操，会安装非常多的包，
需要注意安装位置必须是目标的虚拟环境，
此处为刚新建的虚拟环境，所以进行激活

 conda activate new_langgrapg_env


C:\Users\HP\miniconda3\envs\new_langgrapg_env> conda activate new_langgrapg_env

(new_langgrapg_env) C:\Users\HP\miniconda3\envs\new_langgrapg_env>pip install --upgrade "langgraph-cli[inmem]"
Collecting langgraph-cli[inmem]
  Using cached langgraph_cli-0.4.11-py3-none-any.whl.metadata (4.0 kB)
Collecting click>=8.1.7 (from langgraph-cli[inmem])
  Using cached click-8.3.1-py3-none-any.whl.metadata (2.6 kB)
Collecting langgraph-sdk>=0.1.0 (from langgraph-cli[inmem])
  Using cached langgraph_sdk-0.3.1-py3-none-any.whl.metadata (1.6 kB)
Collecting langgraph-api<0.7.0,>=0.5.35 (from langgraph-cli[inmem])
  Using cached langgraph_api-0.6.15-py3-none-any.whl.metadata (4.2 kB)
Collecting langgraph-runtime-inmem>=0.7 (from langgraph-cli[inmem])
  Using cached langgraph_runtime_inmem-0.20.1-py3-none-any.whl.metadata (570 bytes)
Collecting python-dotenv>=0.8.0 (from langgraph-cli[inmem])
  Using cached python_dotenv-1.2.1-py3-none-any.whl.metadata (25 kB)
Collecting cloudpickle>=3.0.0 (from langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Using cached cloudpickle-3.1.2-py3-none-any.whl.metadata (7.1 kB)
Collecting cryptography<47.0,>=42.0.0 (from langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Using cached cryptography-46.0.3-cp311-abi3-win_amd64.whl.metadata (5.7 kB)
Collecting grpcio-tools==1.75.1 (from langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Downloading grpcio_tools-1.75.1-cp311-cp311-win_amd64.whl.metadata (5.5 kB)
Collecting grpcio<2.0.0,>=1.75.0 (from langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Downloading grpcio-1.76.0-cp311-cp311-win_amd64.whl.metadata (3.8 kB)
Collecting httpx>=0.25.0 (from langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Using cached httpx-0.28.1-py3-none-any.whl.metadata (7.1 kB)
Collecting jsonschema-rs<0.30,>=0.20.0 (from langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Downloading jsonschema_rs-0.29.1-cp311-cp311-win_amd64.whl.metadata (14 kB)
Collecting langchain-core>=0.3.64 (from langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Using cached langchain_core-1.2.5-py3-none-any.whl.metadata (3.7 kB)
Collecting langgraph-checkpoint<5,>=3.0.1 (from langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Using cached langgraph_checkpoint-3.0.1-py3-none-any.whl.metadata (4.7 kB)
Collecting langgraph<2,>=0.4.10 (from langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Using cached langgraph-1.0.5-py3-none-any.whl.metadata (7.4 kB)
Collecting langsmith>=0.3.45 (from langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Using cached langsmith-0.5.1-py3-none-any.whl.metadata (15 kB)
Collecting opentelemetry-api>=1.37.0 (from langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Using cached opentelemetry_api-1.39.1-py3-none-any.whl.metadata (1.5 kB)
Collecting opentelemetry-exporter-otlp-proto-http>=1.37.0 (from langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Using cached opentelemetry_exporter_otlp_proto_http-1.39.1-py3-none-any.whl.metadata (2.4 kB)
Collecting opentelemetry-sdk>=1.37.0 (from langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Using cached opentelemetry_sdk-1.39.1-py3-none-any.whl.metadata (1.5 kB)
Collecting orjson>=3.9.7 (from langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Downloading orjson-3.11.5-cp311-cp311-win_amd64.whl.metadata (42 kB)
Collecting protobuf<7.0.0,>=6.32.1 (from langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Using cached protobuf-6.33.2-cp310-abi3-win_amd64.whl.metadata (593 bytes)
Collecting pyjwt>=2.9.0 (from langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Using cached PyJWT-2.10.1-py3-none-any.whl.metadata (4.0 kB)
Collecting sse-starlette<2.2.0,>=2.1.0 (from langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Using cached sse_starlette-2.1.3-py3-none-any.whl.metadata (5.8 kB)
Collecting starlette>=0.38.6 (from langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Using cached starlette-0.50.0-py3-none-any.whl.metadata (6.3 kB)
Collecting structlog<26,>=24.1.0 (from langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Using cached structlog-25.5.0-py3-none-any.whl.metadata (9.5 kB)
Collecting tenacity>=8.0.0 (from langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Using cached tenacity-9.1.2-py3-none-any.whl.metadata (1.2 kB)
Collecting truststore>=0.1 (from langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Using cached truststore-0.10.4-py3-none-any.whl.metadata (4.4 kB)
Collecting uuid-utils>=0.12.0 (from langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Using cached uuid_utils-0.12.0-cp39-abi3-win_amd64.whl.metadata (1.1 kB)
Collecting uvicorn>=0.26.0 (from langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Using cached uvicorn-0.40.0-py3-none-any.whl.metadata (6.7 kB)
Collecting watchfiles>=0.13 (from langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Downloading watchfiles-1.1.1-cp311-cp311-win_amd64.whl.metadata (5.0 kB)
Requirement already satisfied: setuptools in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from grpcio-tools==1.75.1->langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem]) (80.9.0)
Collecting cffi>=2.0.0 (from cryptography<47.0,>=42.0.0->langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Downloading cffi-2.0.0-cp311-cp311-win_amd64.whl.metadata (2.6 kB)
Collecting typing-extensions~=4.12 (from grpcio<2.0.0,>=1.75.0->langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Using cached typing_extensions-4.15.0-py3-none-any.whl.metadata (3.3 kB)
Collecting langgraph-prebuilt<1.1.0,>=1.0.2 (from langgraph<2,>=0.4.10->langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Using cached langgraph_prebuilt-1.0.5-py3-none-any.whl.metadata (5.2 kB)
Collecting pydantic>=2.7.4 (from langgraph<2,>=0.4.10->langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Using cached pydantic-2.12.5-py3-none-any.whl.metadata (90 kB)
Collecting xxhash>=3.5.0 (from langgraph<2,>=0.4.10->langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Downloading xxhash-3.6.0-cp311-cp311-win_amd64.whl.metadata (13 kB)
Collecting ormsgpack>=1.12.0 (from langgraph-checkpoint<5,>=3.0.1->langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Downloading ormsgpack-1.12.1-cp311-cp311-win_amd64.whl.metadata (3.3 kB)
Collecting blockbuster<2.0.0,>=1.5.24 (from langgraph-runtime-inmem>=0.7->langgraph-cli[inmem])
  Using cached blockbuster-1.5.26-py3-none-any.whl.metadata (10 kB)
Collecting forbiddenfruit>=0.1.4 (from blockbuster<2.0.0,>=1.5.24->langgraph-runtime-inmem>=0.7->langgraph-cli[inmem])
  Using cached forbiddenfruit-0.1.4.tar.gz (43 kB)
  Installing build dependencies ... done
  Getting requirements to build wheel ... done
  Preparing metadata (pyproject.toml) ... done
Collecting anyio (from sse-starlette<2.2.0,>=2.1.0->langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Using cached anyio-4.12.0-py3-none-any.whl.metadata (4.3 kB)
Collecting pycparser (from cffi>=2.0.0->cryptography<47.0,>=42.0.0->langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Using cached pycparser-2.23-py3-none-any.whl.metadata (993 bytes)
Collecting colorama (from click>=8.1.7->langgraph-cli[inmem])
  Using cached colorama-0.4.6-py2.py3-none-any.whl.metadata (17 kB)
Collecting certifi (from httpx>=0.25.0->langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Using cached certifi-2025.11.12-py3-none-any.whl.metadata (2.5 kB)
Collecting httpcore==1.* (from httpx>=0.25.0->langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Using cached httpcore-1.0.9-py3-none-any.whl.metadata (21 kB)
Collecting idna (from httpx>=0.25.0->langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Using cached idna-3.11-py3-none-any.whl.metadata (8.4 kB)
Collecting h11>=0.16 (from httpcore==1.*->httpx>=0.25.0->langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Using cached h11-0.16.0-py3-none-any.whl.metadata (8.3 kB)
Collecting jsonpatch<2.0.0,>=1.33.0 (from langchain-core>=0.3.64->langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Using cached jsonpatch-1.33-py2.py3-none-any.whl.metadata (3.0 kB)
Collecting packaging<26.0.0,>=23.2.0 (from langchain-core>=0.3.64->langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Using cached packaging-25.0-py3-none-any.whl.metadata (3.3 kB)
Collecting pyyaml<7.0.0,>=5.3.0 (from langchain-core>=0.3.64->langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Downloading pyyaml-6.0.3-cp311-cp311-win_amd64.whl.metadata (2.4 kB)
Collecting jsonpointer>=1.9 (from jsonpatch<2.0.0,>=1.33.0->langchain-core>=0.3.64->langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Using cached jsonpointer-3.0.0-py2.py3-none-any.whl.metadata (2.3 kB)
Collecting requests-toolbelt>=1.0.0 (from langsmith>=0.3.45->langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Using cached requests_toolbelt-1.0.0-py2.py3-none-any.whl.metadata (14 kB)
Collecting requests>=2.0.0 (from langsmith>=0.3.45->langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Using cached requests-2.32.5-py3-none-any.whl.metadata (4.9 kB)
Collecting zstandard>=0.23.0 (from langsmith>=0.3.45->langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Downloading zstandard-0.25.0-cp311-cp311-win_amd64.whl.metadata (3.3 kB)
Collecting annotated-types>=0.6.0 (from pydantic>=2.7.4->langgraph<2,>=0.4.10->langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Using cached annotated_types-0.7.0-py3-none-any.whl.metadata (15 kB)
Collecting pydantic-core==2.41.5 (from pydantic>=2.7.4->langgraph<2,>=0.4.10->langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Downloading pydantic_core-2.41.5-cp311-cp311-win_amd64.whl.metadata (7.4 kB)
Collecting typing-inspection>=0.4.2 (from pydantic>=2.7.4->langgraph<2,>=0.4.10->langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Using cached typing_inspection-0.4.2-py3-none-any.whl.metadata (2.6 kB)
Collecting importlib-metadata<8.8.0,>=6.0 (from opentelemetry-api>=1.37.0->langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Using cached importlib_metadata-8.7.1-py3-none-any.whl.metadata (4.7 kB)
Collecting zipp>=3.20 (from importlib-metadata<8.8.0,>=6.0->opentelemetry-api>=1.37.0->langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Using cached zipp-3.23.0-py3-none-any.whl.metadata (3.6 kB)
Collecting googleapis-common-protos~=1.52 (from opentelemetry-exporter-otlp-proto-http>=1.37.0->langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Using cached googleapis_common_protos-1.72.0-py3-none-any.whl.metadata (9.4 kB)
Collecting opentelemetry-exporter-otlp-proto-common==1.39.1 (from opentelemetry-exporter-otlp-proto-http>=1.37.0->langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Using cached opentelemetry_exporter_otlp_proto_common-1.39.1-py3-none-any.whl.metadata (1.8 kB)
Collecting opentelemetry-proto==1.39.1 (from opentelemetry-exporter-otlp-proto-http>=1.37.0->langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Using cached opentelemetry_proto-1.39.1-py3-none-any.whl.metadata (2.3 kB)
Collecting opentelemetry-semantic-conventions==0.60b1 (from opentelemetry-sdk>=1.37.0->langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Using cached opentelemetry_semantic_conventions-0.60b1-py3-none-any.whl.metadata (2.4 kB)
Collecting charset_normalizer<4,>=2 (from requests>=2.0.0->langsmith>=0.3.45->langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Downloading charset_normalizer-3.4.4-cp311-cp311-win_amd64.whl.metadata (38 kB)
Collecting urllib3<3,>=1.21.1 (from requests>=2.0.0->langsmith>=0.3.45->langgraph-api<0.7.0,>=0.5.35->langgraph-cli[inmem])
  Using cached urllib3-2.6.2-py3-none-any.whl.metadata (6.6 kB)
Using cached langgraph_cli-0.4.11-py3-none-any.whl (41 kB)
Using cached langgraph_api-0.6.15-py3-none-any.whl (322 kB)
Downloading grpcio_tools-1.75.1-cp311-cp311-win_amd64.whl (1.2 MB)
   ---------------------------------------- 1.2/1.2 MB 28.9 MB/s  0:00:00
Using cached cryptography-46.0.3-cp311-abi3-win_amd64.whl (3.5 MB)
Downloading grpcio-1.76.0-cp311-cp311-win_amd64.whl (4.7 MB)
   ---------------------------------------- 4.7/4.7 MB 35.6 MB/s  0:00:00
Downloading jsonschema_rs-0.29.1-cp311-cp311-win_amd64.whl (1.9 MB)
   ---------------------------------------- 1.9/1.9 MB 100.6 MB/s  0:00:00
Using cached langgraph-1.0.5-py3-none-any.whl (157 kB)
Using cached langgraph_checkpoint-3.0.1-py3-none-any.whl (46 kB)
Using cached langgraph_prebuilt-1.0.5-py3-none-any.whl (35 kB)
Using cached langgraph_runtime_inmem-0.20.1-py3-none-any.whl (35 kB)
Using cached blockbuster-1.5.26-py3-none-any.whl (13 kB)
Using cached langgraph_sdk-0.3.1-py3-none-any.whl (66 kB)
Using cached protobuf-6.33.2-cp310-abi3-win_amd64.whl (436 kB)
Using cached sse_starlette-2.1.3-py3-none-any.whl (9.4 kB)
Using cached structlog-25.5.0-py3-none-any.whl (72 kB)
Using cached typing_extensions-4.15.0-py3-none-any.whl (44 kB)
Downloading cffi-2.0.0-cp311-cp311-win_amd64.whl (182 kB)
Using cached click-8.3.1-py3-none-any.whl (108 kB)
Using cached cloudpickle-3.1.2-py3-none-any.whl (22 kB)
Using cached httpx-0.28.1-py3-none-any.whl (73 kB)
Using cached httpcore-1.0.9-py3-none-any.whl (78 kB)
Using cached h11-0.16.0-py3-none-any.whl (37 kB)
Using cached langchain_core-1.2.5-py3-none-any.whl (484 kB)
Using cached jsonpatch-1.33-py2.py3-none-any.whl (12 kB)
Using cached langsmith-0.5.1-py3-none-any.whl (275 kB)
Using cached packaging-25.0-py3-none-any.whl (66 kB)
Using cached pydantic-2.12.5-py3-none-any.whl (463 kB)
Downloading pydantic_core-2.41.5-cp311-cp311-win_amd64.whl (2.0 MB)
   ---------------------------------------- 2.0/2.0 MB 116.6 MB/s  0:00:00
Downloading pyyaml-6.0.3-cp311-cp311-win_amd64.whl (158 kB)
Using cached tenacity-9.1.2-py3-none-any.whl (28 kB)
Using cached uuid_utils-0.12.0-cp39-abi3-win_amd64.whl (183 kB)
Using cached annotated_types-0.7.0-py3-none-any.whl (13 kB)
Using cached jsonpointer-3.0.0-py2.py3-none-any.whl (7.6 kB)
Using cached opentelemetry_api-1.39.1-py3-none-any.whl (66 kB)
Using cached importlib_metadata-8.7.1-py3-none-any.whl (27 kB)
Using cached opentelemetry_exporter_otlp_proto_http-1.39.1-py3-none-any.whl (19 kB)
Using cached opentelemetry_exporter_otlp_proto_common-1.39.1-py3-none-any.whl (18 kB)
Using cached opentelemetry_proto-1.39.1-py3-none-any.whl (72 kB)
Using cached googleapis_common_protos-1.72.0-py3-none-any.whl (297 kB)
Using cached opentelemetry_sdk-1.39.1-py3-none-any.whl (132 kB)
Using cached opentelemetry_semantic_conventions-0.60b1-py3-none-any.whl (219 kB)
Using cached requests-2.32.5-py3-none-any.whl (64 kB)
Downloading charset_normalizer-3.4.4-cp311-cp311-win_amd64.whl (106 kB)
Using cached idna-3.11-py3-none-any.whl (71 kB)
Using cached urllib3-2.6.2-py3-none-any.whl (131 kB)
Using cached certifi-2025.11.12-py3-none-any.whl (159 kB)
Downloading orjson-3.11.5-cp311-cp311-win_amd64.whl (133 kB)
Downloading ormsgpack-1.12.1-cp311-cp311-win_amd64.whl (115 kB)
Using cached PyJWT-2.10.1-py3-none-any.whl (22 kB)
Using cached python_dotenv-1.2.1-py3-none-any.whl (21 kB)
Using cached requests_toolbelt-1.0.0-py2.py3-none-any.whl (54 kB)
Using cached starlette-0.50.0-py3-none-any.whl (74 kB)
Using cached anyio-4.12.0-py3-none-any.whl (113 kB)
Using cached truststore-0.10.4-py3-none-any.whl (18 kB)
Using cached typing_inspection-0.4.2-py3-none-any.whl (14 kB)
Using cached uvicorn-0.40.0-py3-none-any.whl (68 kB)
Downloading watchfiles-1.1.1-cp311-cp311-win_amd64.whl (287 kB)
Downloading xxhash-3.6.0-cp311-cp311-win_amd64.whl (31 kB)
Using cached zipp-3.23.0-py3-none-any.whl (10 kB)
Downloading zstandard-0.25.0-cp311-cp311-win_amd64.whl (506 kB)
Using cached colorama-0.4.6-py2.py3-none-any.whl (25 kB)
Using cached pycparser-2.23-py3-none-any.whl (118 kB)
Building wheels for collected packages: forbiddenfruit
  Building wheel for forbiddenfruit (pyproject.toml) ... done
  Created wheel for forbiddenfruit: filename=forbiddenfruit-0.1.4-py3-none-any.whl size=21929 sha256=4fac7d335cb4afa51d74aace99e30eccdeb0d914ac5bb367589eb34bb30b6f7b
  Stored in directory: c:\users\hp\appdata\local\pip\cache\wheels\a3\09\38\5a94e2df4ae8f2cfc8053997507ad46b07009af3afc34a9a9c
Successfully built forbiddenfruit
Installing collected packages: forbiddenfruit, zstandard, zipp, xxhash, uuid-utils, urllib3, typing-extensions, truststore, tenacity, structlog, pyyaml, python-dotenv, pyjwt, pycparser, protobuf, packaging, ormsgpack, orjson, jsonschema-rs, jsonpointer, idna, h11, colorama, cloudpickle, charset_normalizer, certifi, blockbuster, annotated-types, typing-inspection, requests, pydantic-core, opentelemetry-proto, jsonpatch, importlib-metadata, httpcore, grpcio, googleapis-common-protos, click, cffi, anyio, watchfiles, uvicorn, starlette, requests-toolbelt, pydantic, opentelemetry-exporter-otlp-proto-common, opentelemetry-api, httpx, grpcio-tools, cryptography, sse-starlette, opentelemetry-semantic-conventions, langsmith, langgraph-sdk, opentelemetry-sdk, langgraph-cli, langchain-core, opentelemetry-exporter-otlp-proto-http, langgraph-checkpoint, langgraph-prebuilt, langgraph, langgraph-runtime-inmem, langgraph-api
Successfully installed annotated-types-0.7.0 anyio-4.12.0 blockbuster-1.5.26 certifi-2025.11.12 cffi-2.0.0 charset_normalizer-3.4.4 click-8.3.1 cloudpickle-3.1.2 colorama-0.4.6 cryptography-46.0.3 forbiddenfruit-0.1.4 googleapis-common-protos-1.72.0 grpcio-1.76.0 grpcio-tools-1.75.1 h11-0.16.0 httpcore-1.0.9 httpx-0.28.1 idna-3.11 importlib-metadata-8.7.1 jsonpatch-1.33 jsonpointer-3.0.0 jsonschema-rs-0.29.1 langchain-core-1.2.5 langgraph-1.0.5 langgraph-api-0.6.15 langgraph-checkpoint-3.0.1 langgraph-cli-0.4.11 langgraph-prebuilt-1.0.5 langgraph-runtime-inmem-0.20.1 langgraph-sdk-0.3.1 langsmith-0.5.1 opentelemetry-api-1.39.1 opentelemetry-exporter-otlp-proto-common-1.39.1 opentelemetry-exporter-otlp-proto-http-1.39.1 opentelemetry-proto-1.39.1 opentelemetry-sdk-1.39.1 opentelemetry-semantic-conventions-0.60b1 orjson-3.11.5 ormsgpack-1.12.1 packaging-25.0 protobuf-6.33.2 pycparser-2.23 pydantic-2.12.5 pydantic-core-2.41.5 pyjwt-2.10.1 python-dotenv-1.2.1 pyyaml-6.0.3 requests-2.32.5 requests-toolbelt-1.0.0 sse-starlette-2.1.3 starlette-0.50.0 structlog-25.5.0 tenacity-9.1.2 truststore-0.10.4 typing-extensions-4.15.0 typing-inspection-0.4.2 urllib3-2.6.2 uuid-utils-0.12.0 uvicorn-0.40.0 watchfiles-1.1.1 xxhash-3.6.0 zipp-3.23.0 zstandard-0.25.0

(new_langgrapg_env) C:\Users\HP\miniconda3\envs\new_langgrapg_env>
(new_langgrapg_env) C:\Users\HP\miniconda3\envs\new_langgrapg_env>

电脑中实际位置
在这里插入图片描述

也可以查看包


(new_langgrapg_env) C:\Users\HP\miniconda3\envs\new_langgrapg_env>
(new_langgrapg_env) C:\Users\HP\miniconda3\envs\new_langgrapg_env>pip list
Package                                  Version
---------------------------------------- ----------
annotated-types                          0.7.0
anyio                                    4.12.0
blockbuster                              1.5.26
certifi                                  2025.11.12
cffi                                     2.0.0
charset-normalizer                       3.4.4
click                                    8.3.1
cloudpickle                              3.1.2
colorama                                 0.4.6
cryptography                             46.0.3
forbiddenfruit                           0.1.4
googleapis-common-protos                 1.72.0
grpcio                                   1.76.0
grpcio-tools                             1.75.1
h11                                      0.16.0
httpcore                                 1.0.9
httpx                                    0.28.1
idna                                     3.11
importlib_metadata                       8.7.1
jsonpatch                                1.33
jsonpointer                              3.0.0
jsonschema_rs                            0.29.1
langchain-core                           1.2.5
langgraph                                1.0.5
langgraph-api                            0.6.15
langgraph-checkpoint                     3.0.1
langgraph-cli                            0.4.11
langgraph-prebuilt                       1.0.5
langgraph-runtime-inmem                  0.20.1
langgraph-sdk                            0.3.1
langsmith                                0.5.1
opentelemetry-api                        1.39.1
opentelemetry-exporter-otlp-proto-common 1.39.1
opentelemetry-exporter-otlp-proto-http   1.39.1
opentelemetry-proto                      1.39.1
opentelemetry-sdk                        1.39.1
opentelemetry-semantic-conventions       0.60b1
orjson                                   3.11.5
ormsgpack                                1.12.1
packaging                                25.0
pip                                      25.3
protobuf                                 6.33.2
pycparser                                2.23
pydantic                                 2.12.5
pydantic_core                            2.41.5
PyJWT                                    2.10.1
python-dotenv                            1.2.1
PyYAML                                   6.0.3
requests                                 2.32.5
requests-toolbelt                        1.0.0
setuptools                               80.9.0
sse-starlette                            2.1.3
starlette                                0.50.0
structlog                                25.5.0
tenacity                                 9.1.2
truststore                               0.10.4
typing_extensions                        4.15.0
typing-inspection                        0.4.2
urllib3                                  2.6.2
uuid_utils                               0.12.0
uvicorn                                  0.40.0
watchfiles                               1.1.1
wheel                                    0.45.1
xxhash                                   3.6.0
zipp                                     3.23.0
zstandard                                0.25.0

(new_langgrapg_env) C:\Users\HP\miniconda3\envs\new_langgrapg_env>

可以看到 langchain核心包， langgraph ，pip都已经下载

3.创建langgraph-cli应用

实操
注意必须是在虚拟环境里，再进入目标项目存放路径，
再输入命令。

(new_langgrapg_env) C:\Users\HP\miniconda3\envs\new_langgrapg_env>cd C:\Users\HP\PycharmProjects

(new_langgrapg_env) C:\Users\HP\PycharmProjects>langgraph new langgraph_demo

这里
第一个问题，
会让从4个模板中选择一个，
默认回车是，模板1：New LangGraph Project – 带 memory 的最小聊天机器人
第二个问题，让选择语言
语言：Python

(new_langgrapg_env) C:\Users\HP\miniconda3\envs\new_langgrapg_env>cd C:\Users\HP\PycharmProjects

(new_langgrapg_env) C:\Users\HP\PycharmProjects>langgraph new langgraph_demo
🌟 Please select a template:
1. New LangGraph Project - A simple, minimal chatbot with memory.
2. ReAct Agent - A simple agent that can be flexibly extended to many tools.
3. Memory Agent - A ReAct-style agent with an additional tool to store memories for use across conversational threads.
4. Retrieval Agent - An agent that includes a retrieval-based question-answering system.
5. Data-enrichment Agent - An agent that performs web searches and organizes its findings into a structured format.
Enter the number of your template choice (default is 1):

You selected: New LangGraph Project - A simple, minimal chatbot with memory.
Choose language (1 for Python 🐍, 2 for JS/TS 🌐): 1
📥 Attempting to download repository as a ZIP archive...
URL: https://github.com/langchain-ai/new-langgraph-project/archive/refs/heads/main.zip
✅ Downloaded and extracted repository to C:\Users\HP\PycharmProjects\langgraph_demo
🎉 New project created at C:\Users\HP\PycharmProjects\langgraph_demo

(new_langgrapg_env) C:\Users\HP\PycharmProjects>

电脑中成功在虚拟环境中，目标路径下，创建了项目。

在这里插入图片描述
直接用pycharm打开。

总结 CLI 实际完成的事情是：

从 GitHub 下载官方模板

https://github.com/langchain-ai/new-langgraph-project

解压并生成项目到：

C:\Users\HP\PycharmProjects\langgraph_demo

项目创建成功结束（New project created）

说明：

langgraph-cli 安装正常
Python 环境 new_langgrapg_env 正常
网络 / GitHub 访问正常

现在电脑里已经拥有的东西

在这个目录里：

C:\Users\HP\PycharmProjects\langgraph_demo

已经有了一个可运行的 LangGraph 示例项目，通常包含：

基础的 LangGraph 图结构
一个带 memory 的 chatbot
Python 项目结构（可直接用 PyCharm 打开）

5.用 PyCharm 打开创建的项目

在这里插入图片描述

可以手动配置虚拟环境中的3.11版本 python 解释器（也就是创建langgraph项目时候自动下载的pyhon环境）

在这里插入图片描述

6.langgraph项目解析

在这里插入图片描述

项目结构总览

langgraph_demo/
├─ .github/               # CI / GitHub Actions（暂时可忽略）
├─ src/
│  └─ agent/              # ⭐ LangGraph Agent 核心代码
│     ├─ __init__.py
│     └─ graph.py         # ⭐⭐⭐ 最重要
├─ static/                # 前端 / 静态资源（示例项目一般为空）
├─ tests/                 # 测试代码
├─ .env.example           # ⭐⭐⭐环境变量示例（API Key）复制为 .env， ，填入自己的 Key
├─ langgraph.json         # LangGraph CLI / 配置
├─ pyproject.toml         # Python 项目 & 依赖定义
├─ uv.lock                # 依赖锁文件
├─ README.md              # 使用说明
├─ Makefile               # 常用命令封装
└─ LICENSE

非常好，这个截图信息很完整。我按工程视角给你把这个 LangGraph 项目结构逐层拆解，并告诉你真正应该从哪一个文件开始看。

一、整体结论（“导航结论”）

核心入口只有一个文件：

src/agent/graph.py

进项目，90% 的 LangGraph 逻辑都在这里。

如果现在时间有限，只看这个文件就够了。

二、项目结构总览（现在看到的）

langgraph_demo/
├─ .github/               # CI / GitHub Actions（暂时可忽略）
├─ src/
│  └─ agent/              # ⭐ LangGraph Agent 核心代码
│     ├─ __init__.py
│     └─ graph.py         # ⭐⭐⭐ 最重要
├─ static/                # 前端 / 静态资源（示例项目一般为空）
├─ tests/                 # 测试代码
├─ .env.example           # 环境变量示例（API Key）
├─ langgraph.json         # LangGraph CLI / 配置
├─ pyproject.toml         # Python 项目 & 依赖定义
├─ uv.lock                # 依赖锁文件
├─ README.md              # 使用说明
├─ Makefile               # 常用命令封装
└─ LICENSE

三、最应该看的 5 个文件（按顺序）

① `README.md` —— 运行方式 & 官方意图

先看这个，搞清楚：

项目是怎么启动的
需要哪些环境变量
示例是“聊天 / Agent / Memory”哪一类

② `.env.example` —— LLM 配置模板（非常关键）

通常会看到类似：

OPENAI_API_KEY=

需要：

复制为 .env
填入你自己的 Key

③ `pyproject.toml` —— 环境的配置文件:安装项目环境依赖 & Python 版本

依赖项，以便服务器使用您的本地更改。在您的新 LangGraph 应用的根目录下，以编辑模式安类
在LangGraph 中，pyproject.toml代传统的tup.py 和 requirements.txt,可能包含以下扩展配置:

依赖分组:如[project.optional-dependencis]定义 dev(测试框架)依赖。(开发工具)和 test指定Python 版本兼容性。
动态版本控制:通过 requires-python=“>=3
CI/CD 集成:通过[tool.*]配置与 GitHub Actons 或 GitLab CI 的交互

这里可以确认：

用的是哪个 Python 版本
是否依赖 langgraph / langchain
是否使用 uv

现在用的是 conda + pip，这没问题。

[project]
# 项目名称（Python 包名）
# pip install -e . 时使用的名字
name = "agent"

# 项目版本号
version = "0.0.1"

# 项目描述（用于包说明、README）
description = "Starter template for making a new agent LangGraph."

# 作者信息（基本不影响运行）
authors = [
    { name = "William Fu-Hinthorn", email = "13333726+hinthornw@users.noreply.github.com" },
]

# 项目说明文档
readme = "README.md"

# 开源协议
license = { text = "MIT" }

# Python 版本要求（非常重要）
# 说明：必须使用 Python 3.10 或以上
requires-python = ">=3.10"

# 运行时必须安装的依赖（核心）
dependencies = [
    "langgraph>=1.0.0",        # LangGraph 核心框架
    "python-dotenv>=1.0.1",    # 用于加载 .env 环境变量
]

[project.optional-dependencies]
# 开发环境专用依赖
# 一般不会部署到生产
dev = [
    "mypy>=1.11.1",   # 静态类型检查
    "ruff>=0.6.1",    # 代码格式 / Lint 工具
]

CLI 相关
说明：

当前项目是为 LangGraph CLI 服务的

使用的是 in-memory 运行模式

对应之前执行的命令：

langgraph new langgraph_demo

[dependency-groups]
dev = [
    "anyio>=4.7.0",
    "langgraph-cli[inmem]>=0.4.7",
    "mypy>=1.13.0",
    "pytest>=8.3.5",
    "ruff>=0.8.2",
]

④ `langgraph.json` —— LangGraph 项目的配置文件

这个文件是 LangGraph CLI 识别项目的关键，通常包含：

Agent 定义
启动入口
CLI / Studio 相关配置

一般不用手改，但知道它的存在很重要。

{
  "$schema": "https://langgra.ph/schema.json",
  "dependencies": ["."],
  "graphs": {
    "agent": "./src/agent/graph.py:graph"
  },
  "env": ".env",
  "image_distro": "wolfi"
}

{
  // 运行 Graph 时需要安装的 Python 依赖
  // "." 表示：安装当前项目根目录本身（即 pyproject.toml）
  // 等价于：pip install -e .
  "dependencies": ["."],

  // 定义 Graph 的入口
  // key: Graph 名称（CLI / Studio 中使用）
  // value: 智能体对象。Python 文件路径 : 变量名
  "graphs": {
    "agent": "./src/agent/graph.py:graph"
  },

  // 启动时自动加载的环境变量文件
  // 一般用于 OPENAI_API_KEY 等
  "env": ".env",

  // 构建容器 / 远程运行时使用的基础镜像发行版 docker等
  // wolfi = 官方推荐、安全性高
  // 本地开发阶段基本不用管
  "image_distro": "wolfi"
}

⑤ ⭐ `src/agent/graph.py` —— 真正的“业务逻辑”

这就是要精读的文件

这里通常会看到：

State 定义（类似 Java DTO）
Node（节点函数）
Graph（节点如何连接）
Memory / Checkpoint 机制

四、用 Java / 后端工程的角度理解（很重要）

可以这样类比：

LangGraph	Java 后端类比
State	DTO / VO
Node	Service 方法
Edge	流程控制（if / switch）
Graph	工作流 / 状态机
Memory	Session / Redis / DB

LangGraph 本质 = LLM + 状态机

这也是它比 LangChain 更适合工程化的原因。

这个比较简单，但是进阶，可以：

把它改成 REST API（FastAPI）
加一个 Tool / DB / HTTP 调用
做成 多 Agent 流程

7. pip install -e . 读取 pyproject.toml在虚拟环境安装依赖

当前目录的 Python 项目”以「可编辑模式」安装到当前虚拟环境中

pip install -e .

说明：

pip install -e .

作用

把当前目录的项目安装到当前虚拟环境
使用 可编辑模式（-e），修改源码后立刻生效

具体做了什么

读取 pyproject.toml
安装依赖（如 langgraph、python-dotenv）
让我们可以直接 import agent

什么时候需要再执行

改代码：不需要
改 pyproject.toml / 依赖：需要

Microsoft Windows [版本 10.0.19045.6466]
(c) Microsoft Corporation。保留所有权利。

(new_langgrapg_env) C:\Users\HP\PycharmProjects\langgraph_demo>pip install -e . 
Obtaining file:///C:/Users/HP/PycharmProjects/langgraph_demo
  Installing build dependencies ... done
  Checking if build backend supports build_editable ... done
  Getting requirements to build editable ... done
  Preparing editable metadata (pyproject.toml) ... done
Requirement already satisfied: langgraph>=1.0.0 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from agent==0.0.1) (1.0.5)
Requirement already satisfied: python-dotenv>=1.0.1 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from agent==0.0.1) (1.2.1)
Requirement already satisfied: langchain-core>=0.1 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from langgraph>=1.0.0->agent==0.0.1) (1.2.5)
Requirement already satisfied: langgraph-checkpoint<4.0.0,>=2.1.0 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from langgraph>=1.0.0->agent==0.0.1) (3.0.1)
Requirement already satisfied: langgraph-prebuilt<1.1.0,>=1.0.2 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from langgraph>=1.0.0->agent==0.0.1) (1.0.5)
Requirement already satisfied: langgraph-sdk<0.4.0,>=0.3.0 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from langgraph>=1.0.0->agent==0.0.1) (0.3.1)
Requirement already satisfied: pydantic>=2.7.4 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from langgraph>=1.0.0->agent==0.0.1) (2.12.5)
Requirement already satisfied: xxhash>=3.5.0 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from langgraph>=1.0.0->agent==0.0.1) (3.6.0)
Requirement already satisfied: ormsgpack>=1.12.0 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from langgraph-checkpoint<4.0.0,>=2.1.0->langgraph>=1.0.0->agent==0.0.1) (1.12.1)
Requirement already satisfied: httpx>=0.25.2 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from langgraph-sdk<0.4.0,>=0.3.0->langgraph>=1.0.0->agent==0.0.1) (0.28.1)
Requirement already satisfied: orjson>=3.10.1 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from langgraph-sdk<0.4.0,>=0.3.0->langgraph>=1.0.0->agent==0.0.1) (3.11.5)
Requirement already satisfied: anyio in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from httpx>=0.25.2->langgraph-sdk<0.4.0,>=0.3.0->langgraph>=1.0.0->agent==0.0.1) (4.12.0)
Requirement already satisfied: certifi in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from httpx>=0.25.2->langgraph-sdk<0.4.0,>=0.3.0->langgraph>=1.0.0->agent==0.0.1) (2025.11.12)
Requirement already satisfied: httpcore==1.* in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from httpx>=0.25.2->langgraph-sdk<0.4.0,>=0.3.0->langgraph>=1.0.0->agent==0.0.1) (1.0.9)
Requirement already satisfied: idna in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from httpx>=0.25.2->langgraph-sdk<0.4.0,>=0.3.0->langgraph>=1.0.0->agent==0.0.1) (3.11)
Requirement already satisfied: h11>=0.16 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from httpcore==1.*->httpx>=0.25.2->langgraph-sdk<0.4.0,>=0.3.0->langgraph>=1.0.0->agent==0.0.1) (0.16.0)
Requirement already satisfied: jsonpatch<2.0.0,>=1.33.0 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from langchain-core>=0.1->langgraph>=1.0.0->agent==0.0.1) (1.33)
Requirement already satisfied: langsmith<1.0.0,>=0.3.45 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from langchain-core>=0.1->langgraph>=1.0.0->agent==0.0.1) (0.5.1)
Requirement already satisfied: packaging<26.0.0,>=23.2.0 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from langchain-core>=0.1->langgraph>=1.0.0->agent==0.0.1) (25.0)
Requirement already satisfied: pyyaml<7.0.0,>=5.3.0 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from langchain-core>=0.1->langgraph>=1.0.0->agent==0.0.1) (6.0.3)
Requirement already satisfied: tenacity!=8.4.0,<10.0.0,>=8.1.0 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from langchain-core>=0.1->langgraph>=1.0.0->agent==0.0.1) (9.1.2)
Requirement already satisfied: typing-extensions<5.0.0,>=4.7.0 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from langchain-core>=0.1->langgraph>=1.0.0->agent==0.0.1) (4.15.0)
Requirement already satisfied: uuid-utils<1.0,>=0.12.0 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from langchain-core>=0.1->langgraph>=1.0.0->agent==0.0.1) (0.12.0)
Requirement already satisfied: jsonpointer>=1.9 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from jsonpatch<2.0.0,>=1.33.0->langchain-core>=0.1->langgraph>=1.0.0->agent==0.0.1) (3.0.0)
Requirement already satisfied: requests-toolbelt>=1.0.0 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from langsmith<1.0.0,>=0.3.45->langchain-core>=0.1->langgraph>=1.0.0->agent==0.0.1) (1.0.0)
Requirement already satisfied: requests>=2.0.0 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from langsmith<1.0.0,>=0.3.45->langchain-core>=0.1->langgraph>=1.0.0->agent==0.0.1) (2.32.5)
Requirement already satisfied: zstandard>=0.23.0 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from langsmith<1.0.0,>=0.3.45->langchain-core>=0.1->langgraph>=1.0.0->agent==0.0.1) (0.25.0)
Requirement already satisfied: annotated-types>=0.6.0 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from pydantic>=2.7.4->langgraph>=1.0.0->agent==0.0.1) (0.7.0)
Requirement already satisfied: pydantic-core==2.41.5 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from pydantic>=2.7.4->langgraph>=1.0.0->agent==0.0.1) (2.41.5)
Requirement already satisfied: typing-inspection>=0.4.2 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from pydantic>=2.7.4->langgraph>=1.0.0->agent==0.0.1) (0.4.2)
Requirement already satisfied: charset_normalizer<4,>=2 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from requests>=2.0.0->langsmith<1.0.0,>=0.3.45->langchain-core>=0.1->langgraph>=1.0.0->agent==0.0.1) (3.4.4)
Requirement already satisfied: urllib3<3,>=1.21.1 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from requests>=2.0.0->langsmith<1.0.0,>=0.3.45->langchain-core>=0.1->langgraph>=1.0.0->agent==0.0.1) (2.6.2)
Building wheels for collected packages: agent
  Building editable for agent (pyproject.toml) ... done
  Created wheel for agent: filename=agent-0.0.1-0.editable-py3-none-any.whl size=5144 sha256=a20865e802be5e6ebdd07270ad3ffcb8bc365c47d40879ff204dc7e1ada9c7af
  Stored in directory: C:\Users\HP\AppData\Local\Temp\pip-ephem-wheel-cache-ayffcelk\wheels\ff\f9\32\fde3f2c0f36f1489ceb5a0b8240e369fdb64b13393b9bb691e
Successfully built agent
Installing collected packages: agent
Successfully installed agent-0.0.1

(new_langgrapg_env) C:\Users\HP\PycharmProjects\langgraph_demo>

8. .env配置环境文件：API KEY安装 llm对应的 LangChain Provider

LLM 为 Gemini 2.5 Flash
LangChain 官方明确 Gemini 需要安装 langchain-google-genai 并用 GOOGLE_API_KEY/GEMINI_API_KEY 等方式鉴权

在这里插入图片描述

pip install -U langchain-google-genai

实操


(new_langgrapg_env) C:\Users\HP\PycharmProjects\langgraph_demo>pip install -U langchain-google-genai
Collecting langchain-google-genai
  Downloading langchain_google_genai-4.1.2-py3-none-any.whl.metadata (2.7 kB)
Collecting filetype<2.0.0,>=1.2.0 (from langchain-google-genai)
  Downloading filetype-1.2.0-py2.py3-none-any.whl.metadata (6.5 kB)
Collecting google-genai<2.0.0,>=1.56.0 (from langchain-google-genai)
  Using cached google_genai-1.56.0-py3-none-any.whl.metadata (53 kB)
Requirement already satisfied: langchain-core<2.0.0,>=1.2.2 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from langchain-google-genai) (1.2.5)
Requirement already satisfied: pydantic<3.0.0,>=2.0.0 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from langchain-google-genai) (2.12.5)
Requirement already satisfied: anyio<5.0.0,>=4.8.0 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from google-genai<2.0.0,>=1.56.0->langchain-google-genai) (4.12.0)
Collecting google-auth<3.0.0,>=2.45.0 (from google-auth[requests]<3.0.0,>=2.45.0->google-genai<2.0.0,>=1.56.0->langchain-google-genai)
  Using cached google_auth-2.45.0-py2.py3-none-any.whl.metadata (6.8 kB)
Requirement already satisfied: httpx<1.0.0,>=0.28.1 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from google-genai<2.0.0,>=1.56.0->langchain-google-genai) (0.28.1)
Requirement already satisfied: requests<3.0.0,>=2.28.1 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from google-genai<2.0.0,>=1.56.0->langchain-google-genai) (2.32.5)
Requirement already satisfied: tenacity<9.2.0,>=8.2.3 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from google-genai<2.0.0,>=1.56.0->langchain-google-genai) (9.1.2)
Collecting websockets<15.1.0,>=13.0.0 (from google-genai<2.0.0,>=1.56.0->langchain-google-genai)
  Downloading websockets-15.0.1-cp311-cp311-win_amd64.whl.metadata (7.0 kB)
Requirement already satisfied: typing-extensions<5.0.0,>=4.11.0 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from google-genai<2.0.0,>=1.56.0->langchain-google-genai) (4.15.0)
Collecting distro<2,>=1.7.0 (from google-genai<2.0.0,>=1.56.0->langchain-google-genai)
  Using cached distro-1.9.0-py3-none-any.whl.metadata (6.8 kB)
Collecting sniffio (from google-genai<2.0.0,>=1.56.0->langchain-google-genai)
  Using cached sniffio-1.3.1-py3-none-any.whl.metadata (3.9 kB)
Requirement already satisfied: idna>=2.8 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from anyio<5.0.0,>=4.8.0->google-genai<2.0.0,>=1.56.0->langchain-google-genai) (3.11)
Collecting cachetools<7.0,>=2.0.0 (from google-auth<3.0.0,>=2.45.0->google-auth[requests]<3.0.0,>=2.45.0->google-genai<2.0.0,>=1.56.0->langchain-google-genai)
  Using cached cachetools-6.2.4-py3-none-any.whl.metadata (5.6 kB)
Collecting pyasn1-modules>=0.2.1 (from google-auth<3.0.0,>=2.45.0->google-auth[requests]<3.0.0,>=2.45.0->google-genai<2.0.0,>=1.56.0->langchain-google-genai)
  Using cached pyasn1_modules-0.4.2-py3-none-any.whl.metadata (3.5 kB)
Collecting rsa<5,>=3.1.4 (from google-auth<3.0.0,>=2.45.0->google-auth[requests]<3.0.0,>=2.45.0->google-genai<2.0.0,>=1.56.0->langchain-google-genai)
  Using cached rsa-4.9.1-py3-none-any.whl.metadata (5.6 kB)
Requirement already satisfied: certifi in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from httpx<1.0.0,>=0.28.1->google-genai<2.0.0,>=1.56.0->langchain-google-genai) (2025.11.12)
Requirement already satisfied: httpcore==1.* in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from httpx<1.0.0,>=0.28.1->google-genai<2.0.0,>=1.56.0->langchain-google-genai) (1.0.9)
Requirement already satisfied: h11>=0.16 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from httpcore==1.*->httpx<1.0.0,>=0.28.1->google-genai<2.0.0,>=1.56.0->langchain-google-genai) (0.16.0)
Requirement already satisfied: jsonpatch<2.0.0,>=1.33.0 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from langchain-core<2.0.0,>=1.2.2->langchain-google-genai) (1.33)
Requirement already satisfied: langsmith<1.0.0,>=0.3.45 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from langchain-core<2.0.0,>=1.2.2->langchain-google-genai) (0.5.1)
Requirement already satisfied: packaging<26.0.0,>=23.2.0 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from langchain-core<2.0.0,>=1.2.2->langchain-google-genai) (25.0)
Requirement already satisfied: pyyaml<7.0.0,>=5.3.0 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from langchain-core<2.0.0,>=1.2.2->langchain-google-genai) (6.0.3)
Requirement already satisfied: uuid-utils<1.0,>=0.12.0 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from langchain-core<2.0.0,>=1.2.2->langchain-google-genai) (0.12.0)
Requirement already satisfied: jsonpointer>=1.9 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from jsonpatch<2.0.0,>=1.33.0->langchain-core<2.0.0,>=1.2.2->langchain-google-genai) (3.0.0)
Requirement already satisfied: orjson>=3.9.14 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from langsmith<1.0.0,>=0.3.45->langchain-core<2.0.0,>=1.2.2->langchain-google-genai) (3.11.5)
Requirement already satisfied: requests-toolbelt>=1.0.0 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from langsmith<1.0.0,>=0.3.45->langchain-core<2.0.0,>=1.2.2->langchain-google-genai) (1.0.0)
Requirement already satisfied: zstandard>=0.23.0 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from langsmith<1.0.0,>=0.3.45->langchain-core<2.0.0,>=1.2.2->langchain-google-genai) (0.25.0)
Requirement already satisfied: annotated-types>=0.6.0 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from pydantic<3.0.0,>=2.0.0->langchain-google-genai) (0.7.0)
Requirement already satisfied: pydantic-core==2.41.5 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from pydantic<3.0.0,>=2.0.0->langchain-google-genai) (2.41.5)
Requirement already satisfied: typing-inspection>=0.4.2 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from pydantic<3.0.0,>=2.0.0->langchain-google-genai) (0.4.2)
Requirement already satisfied: charset_normalizer<4,>=2 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from requests<3.0.0,>=2.28.1->google-genai<2.0.0,>=1.56.0->langchain-google-genai) (3.4.4)
Requirement already satisfied: urllib3<3,>=1.21.1 in c:\users\hp\miniconda3\envs\new_langgrapg_env\lib\site-packages (from requests<3.0.0,>=2.28.1->google-genai<2.0.0,>=1.56.0->langchain-google-genai) (2.6.2)
Collecting pyasn1>=0.1.3 (from rsa<5,>=3.1.4->google-auth<3.0.0,>=2.45.0->google-auth[requests]<3.0.0,>=2.45.0->google-genai<2.0.0,>=1.56.0->langchain-google-genai)
  Using cached pyasn1-0.6.1-py3-none-any.whl.metadata (8.4 kB)
Downloading langchain_google_genai-4.1.2-py3-none-any.whl (65 kB)
Downloading filetype-1.2.0-py2.py3-none-any.whl (19 kB)
Using cached google_genai-1.56.0-py3-none-any.whl (426 kB)
Using cached distro-1.9.0-py3-none-any.whl (20 kB)
Using cached google_auth-2.45.0-py2.py3-none-any.whl (233 kB)
Using cached cachetools-6.2.4-py3-none-any.whl (11 kB)
Using cached rsa-4.9.1-py3-none-any.whl (34 kB)
Downloading websockets-15.0.1-cp311-cp311-win_amd64.whl (176 kB)
Using cached pyasn1-0.6.1-py3-none-any.whl (83 kB)
Using cached pyasn1_modules-0.4.2-py3-none-any.whl (181 kB)
Using cached sniffio-1.3.1-py3-none-any.whl (10 kB)
Installing collected packages: filetype, websockets, sniffio, pyasn1, distro, cachetools, rsa, pyasn1-modules, google-auth, google-genai, langchain-google-genai
Successfully installed cachetools-6.2.4 distro-1.9.0 filetype-1.2.0 google-auth-2.45.0 google-genai-1.56.0 langchain-google-genai-4.1.2 pyasn1-0.6.1 pyasn1-modules-0.4.2 rsa-4.9.1 sniffio-1.3.1 websockets-15.0.1

(new_langgrapg_env) C:\Users\HP\PycharmProjects\langgraph_demo>

然后就可以配置环境文件 .env

在这里插入图片描述

test_gemini.py 项目目录下，可以新预先看llm能否跑通

9.graph.py：AI Agent智能体的业务代码

项目中的位置
在这里插入图片描述

# ============================================================
# 修改前：本地私有化部署的大模型（OpenAI 协议兼容）
# ============================================================

# from langchain_openai import ChatOpenAI
#
# llm = ChatOpenAI(
#     model="qwen3-8b",                 # 使用的模型名称
#     temperature=0.8,                  # 随机性，越大回答越发散
#     api_key="xx",                     # API Key（本地一般随便填）
#     base_url="http://localhost:6006/v1",  # 本地模型的 OpenAI 接口地址
#     extra_body={                      # 额外参数，透传给模型
#         "chat_template_kwargs": {
#             "enable_thinking": False  # 关闭“思考过程”输出（Qwen 模板参数）
#         }
#     },
# )

# ============================================================
# 修改后：Gemini 2.5 Flash（官方 Gemini API + LangChain 适配）
# ============================================================

from langchain_google_genai import ChatGoogleGenerativeAI

from dotenv import load_dotenv

load_dotenv()

import os

print(os.getenv("GOOGLE_API_KEY"))


llm = ChatGoogleGenerativeAI(
    model="gemini-2.5-flash",  # Gemini 2.5 Flash（注意：flash，不是 fash）
    temperature=0.8,  # 含义与 OpenAI / Qwen 完全一致
    thinking_budget=0,  # ⭐ 等价于 enable_thinking=False（关闭 thinking）
)


# ============================================================
# 工具函数（这一部分：前后完全一样，不需要改）
# ============================================================

def get_weather(city: str) -> str:
    """Get weather for a given city."""
    return f"城市 in {city}，今天天气晴朗!"


# ============================================================
# 创建 ReAct Agent（逻辑不变，只是 LLM 实现变了）
# ============================================================

from langgraph.prebuilt import create_react_agent

# 此处的变量名 graph对应langgraph.json     "agent": "./src/agent/graph.py:graph"
graph = create_react_agent(
    llm,  # Gemini 2.5 Flash
    tools=[get_weather],  # Agent 可调用的工具列表
    prompt="你是一个智能助手！"
)

10.langgraph dev启动 LangGraph 开发服务器

langgraph dev

作用

以内存模式（in-memory）启动 LangGraph 服务
适合开发 / 调试
不适合生产（数据不持久化）

二、默认行为（你不用额外配置）

配置文件：langgraph.json
监听地址：127.0.0.1
端口：2024
自动热重载：开启
启动后会自动打开浏览器（如未禁用）

启动成功后一般访问：

http://127.0.0.1:2024

三、常用参数（记这几个就够）

langgraph dev --host 0.0.0.0 --port 2024

参数	类型	默认	说明
`--config`	path	`langgraph.json`	配置文件
`--host`	str	`127.0.0.1`	监听地址
`--port`	int	`2024`	监听端口
`--no-reload`	flag	false	关闭热重载
`--no-browser`	flag	false	不打开浏览器
`--debug-port`	int	-	调试端口
`--wait-for-client`	flag	false	等调试器
`--n-jobs-per-worker`	int	-	并发任务数

四、之后每次项目启动的正确使用顺序（推荐）

# 1. 进入项目目录
cd langgraph_demo

# 2. 确认已安装项目
pip install -e .

# 3. 启动开发服务器
langgraph dev

一句话总结

langgraph dev = 启动本地 LangGraph 开发服务
自动读取 langgraph.json，加载自己定义的 graph

实操

Microsoft Windows [版本 10.0.19045.6466]
(c) Microsoft Corporati



on。保留所有权利。

(new_langgrapg_env) C:\Users\HP\PycharmProjects\langgraph_demo>langgraph dev
INFO:langgraph_api.cli:

        Welcome to

╦  ┌─┐┌┐┌┌─┐╔═╗┬─┐┌─┐┌─┐┬ ┬
║  ├─┤││││ ┬║ ╦├┬┘├─┤├─┘├─┤
╩═╝┴ ┴┘└┘└─┘╚═╝┴└─┴ ┴┴  ┴ ┴

- 🚀 API: http://127.0.0.1:2024
- 🎨 Studio UI: https://smith.langchain.com/studio/?baseUrl=http://127.0.0.1:2024
- 📚 API Docs: http://127.0.0.1:2024/docs

This in-memory server is designed for development and testing.
For production use, please use LangSmith Deployment.


2025-12-29T09:15:24.063012Z [info     ] Using langgraph_runtime_inmem  [langgraph_runtime] api_variant=local_dev langgraph_api_version=0.6.15 thread_name=MainThread
2025-12-29T09:15:24.083068Z [info     ] Using auth of type=noop        [langgraph_api.auth.middleware] api_variant=local_dev langgraph_api_version=0.6.15 thread_name=MainThread
2025-12-29T09:15:24.090068Z [info     ] Starting In-Memory runtime with langgraph-api=0.6.15 and in-memory runtime=0.20.1 [langgraph_runtime_inmem.lifespan] api_variant=local_dev langgraph_api_version=0.6.15 langgraph_runtime_inmem_version=0.20.1 thread_name=asyncio_0 version=0.6.15
2025-12-29T09:15:24.431328Z [info     ] Starting metadata loop         [langgraph_api.metadata] api_variant=local_dev endpoint=https://api.smith.langchain.com/v1/metadata/submit langgraph_api_version=0.6.15 thread_name=MainThread
2025-12-29T09:15:24.432255Z [info     ] Getting auth instance: None    [langgraph_api.auth.custom] api_variant=local_dev langgraph_api_version=0.6.15 langgraph_auth=None thread_name=MainThread
2025-12-29T09:15:24.439808Z [info     ] Starting thread TTL sweeper with interval 5 minutes [langgraph_api.thread_ttl] api_variant=local_dev interval_minutes=5 langgraph_api_version=0.6.15 strategy=delete thread_name=asyncio_1
2025-12-29T09:15:24.673704Z [info     ] HTTP Request: POST https://api.smith.langchain.com/v1/metadata/submit "HTTP/1.1 204 No Content" [httpx] api_variant=local_dev langgraph_api_version=0.6.15 thread_name=MainThread
2025-12-29T09:15:24.720407Z [info     ] Successfully submitted metadata to LangSmith instance [langgraph_api.metadata] api_variant=local_dev langgraph_api_version=0.6.15 n_nodes=0 n_runs=0 thread_name=MainThread
2025-12-29T09:15:25.977287Z [info     ] Importing graph with id agent  [langgraph_api.timing.timer] api_variant=local_dev elapsed_seconds=1.5456026000028942 graph_id=agent langgraph_api_version=0.6.15 module=None name=_graph_from_spec path=./src/agent/graph.py thread_name=asyncio_0
2025-12-29T09:15:25.978291Z [info     ] Application started up in 2.250s [langgraph_api.timing.timer] api_variant=local_dev elapsed=2.25 langgraph_api_version=0.6.15 thread_name=MainThread
2025-12-29T09:15:26.005472Z [info     ] Starting 1 background workers  [langgraph_runtime_inmem.queue] api_variant=local_dev langgraph_api_version=0.6.15 thread_name=asyncio_2
2025-12-29T09:15:26.006576Z [info     ] Worker stats                   [langgraph_runtime_inmem.queue] active=0 api_variant=local_dev available=1 langgraph_api_version=0.6.15 max=1 thread_name=asyncio_0
Server started in 3.74s
2025-12-29T09:15:26.008285Z [info     ] Server started in 3.74s        [browser_opener] api_variant=local_dev langgraph_api_version=0.6.15 message='Server started in 3.74s' thread_name='Thread-2 (_open_browser)'
🎨 Opening Studio in your browser...
2025-12-29T09:15:26.009281Z [info     ] 🎨 Opening Studio in your browser... [browser_opener] api_variant=local_dev langgraph_api_version=0.6.15 message='🎨 Opening Studio in your browser...' thread_name='Thread-2 (_open_browser)'
URL: https://smith.langchain.com/studio/?baseUrl=http://127.0.0.1:2024
2025-12-29T09:15:26.009281Z [info     ] URL: https://smith.langchain.com/studio/?baseUrl=http://127.0.0.1:2024 [browser_opener] api_variant=local_dev langgraph_api_version=0.6.15 message='URL: https://smith.langchain.com/studio/?baseUrl=http://127.0.0.1:2024' thread_name='Thread-2 (_open_browser)'
2025-12-29T09:15:26.520774Z [info     ] Queue stats                    [langgraph_runtime_inmem.queue] api_variant=local_dev langgraph_api_version=0.6.15 n_pending=0 n_running=0 pending_runs_wait_time_max_secs=None pending_runs_wait_time_med_secs=None thread_name=asyncio_2
2025-12-29T09:16:26.230347Z [info     ] Worker stats                   [langgraph_runtime_inmem.queue] active=0 api_variant=local_dev available=1 langgraph_api_version=0.6.15 max=1 thread_name=asyncio_0
2025-12-29T09:16:26.746978Z [info     ] Queue stats                    [langgraph_runtime_inmem.queue] api_variant=local_dev langgraph_api_version=0.6.15 n_pending=0 n_running=0 pending_runs_wait_time_max_secs=None pending_runs_wait_time_med_secs=None thread_name=asyncio_2
2025-12-29T09:17:26.402147Z [info     ] Worker stats                   [langgraph_runtime_inmem.queue] active=0 api_variant=local_dev available=1 langgraph_api_version=0.6.15 max=1 thread_name=asyncio_0
2025-12-29T09:17:26.916070Z [info     ] Queue stats                    [langgraph_runtime_inmem.queue] api_variant=local_dev langgraph_api_version=0.6.15 n_pending=0 n_running=0 pending_runs_wait_time_max_secs=None pending_runs_wait_time_med_secs=None thread_name=asyncio_2
2025-12-29T09:18:26.507434Z [info     ] Worker stats                   [langgraph_runtime_inmem.queue] active=0 api_variant=local_dev available=1 langgraph_api_version=0.6.15 max=1 thread_name=asyncio_0
2025-12-29T09:18:27.007509Z [info     ] Queue stats                    [langgraph_runtime_inmem.queue] api_variant=local_dev langgraph_api_version=0.6.15 n_pending=0 n_running=0 pending_runs_wait_time_max_secs=None pending_runs_wait_time_med_secs=None thread_name=asyncio_2
2025-12-29T09:19:26.606802Z [info     ] Worker stats                   [langgraph_runtime_inmem.queue] active=0 api_variant=local_dev available=1 langgraph_api_version=0.6.15 max=1 thread_name=asyncio_0
2025-12-29T09:19:27.121037Z [info     ] Queue stats                    [langgraph_runtime_inmem.queue] api_variant=local_dev langgraph_api_version=0.6.15 n_pending=0 n_running=0 pending_runs_wait_time_max_secs=None pending_runs_wait_time_med_secs=None thread_name=asyncio_2
2025-12-29T09:19:37.599623Z [info     ] 4 changes detected             [watchfiles.main] api_variant=local_dev langgraph_api_version=0.6.15 thread_name=MainThread
2025-12-29T09:19:38.316082Z [info     ] 1 change detected              [watchfiles.main] api_variant=local_dev langgraph_api_version=0.6.15 thread_name=MainThread
2025-12-29T09:20:24.950454Z [info     ] HTTP Request: POST https://api.smith.langchain.com/v1/metadata/submit "HTTP/1.1 204 No Content" [httpx] api_variant=local_dev langgraph_api_version=0.6.15 thread_name=MainThread
2025-12-29T09:20:24.953441Z [info     ] Successfully submitted metadata to LangSmith instance [langgraph_api.metadata] api_variant=local_dev langgraph_api_version=0.6.15 n_nodes=0 n_runs=0 thread_name=MainThread
2025-12-29T09:20:26.805065Z [info     ] Worker stats                   [langgraph_runtime_inmem.queue] active=0 api_variant=local_dev available=1 langgraph_api_version=0.6.15 max=1 thread_name=asyncio_2
2025-12-29T09:20:27.306354Z [info     ] Queue stats                    [langgraph_runtime_inmem.queue] api_variant=local_dev langgraph_api_version=0.6.15 n_pending=0 n_running=0 pending_runs_wait_time_max_secs=None pending_runs_wait_time_med_secs=None thread_name=asyncio_0
2025-12-29T09:21:27.089428Z [info     ] Worker stats                   [langgraph_runtime_inmem.queue] active=0 api_variant=local_dev available=1 langgraph_api_version=0.6.15 max=1 thread_name=asyncio_2
2025-12-29T09:21:27.588381Z [info     ] Queue stats                    [langgraph_runtime_inmem.queue] api_variant=local_dev langgraph_api_version=0.6.15 n_pending=0 n_running=0 pending_runs_wait_time_max_secs=None pending_runs_wait_time_med_secs=None thread_name=asyncio_0
2025-12-29T09:22:27.178610Z [info     ] Worker stats                   [langgraph_runtime_inmem.queue] active=0 api_variant=local_dev available=1 langgraph_api_version=0.6.15 max=1 thread_name=asyncio_2
2025-12-29T09:22:27.682870Z [info     ] Queue stats                    [langgraph_runtime_inmem.queue] api_variant=local_dev langgraph_api_version=0.6.15 n_pending=0 n_running=0 pending_runs_wait_time_max_secs=None pending_runs_wait_time_med_secs=None thread_name=asyncio_0
2025-12-29T09:23:27.454220Z [info     ] Worker stats                   [langgraph_runtime_inmem.queue] active=0 api_variant=local_dev available=1 langgraph_api_version=0.6.15 max=1 thread_name=asyncio_2
2025-12-29T09:23:27.952285Z [info     ] Queue stats                    [langgraph_runtime_inmem.queue] api_variant=local_dev langgraph_api_version=0.6.15 n_pending=0 n_running=0 pending_runs_wait_time_max_secs=None pending_runs_wait_time_med_secs=None thread_name=asyncio_0
2025-12-29T09:24:27.504043Z [info     ] Worker stats                   [langgraph_runtime_inmem.queue] active=0 api_variant=local_dev available=1 langgraph_api_version=0.6.15 max=1 thread_name=asyncio_2
2025-12-29T09:24:28.019370Z [info     ] Queue stats                    [langgraph_runtime_inmem.queue] api_variant=local_dev langgraph_api_version=0.6.15 n_pending=0 n_running=0 pending_runs_wait_time_max_secs=None pending_runs_wait_time_med_secs=None thread_name=asyncio_0
2025-12-29T09:25:04.898115Z [info     ] 4 changes detected             [watchfiles.main] api_variant=local_dev langgraph_api_version=0.6.15 thread_name=MainThread
2025-12-29T09:25:05.252553Z [info     ] 1 change detected              [watchfiles.main] api_variant=local_dev langgraph_api_version=0.6.15 thread_name=MainThread

11.langSmith 使用

项目成功启动，自动打开langSmith
在这里插入图片描述

在这里插入图片描述

下面按“现在正在用 LangGraph Studio 的真实视角”，把这张图一次性讲清楚

一、这张图是什么？（一句话）

这是当前 Agent 的 LangGraph 执行图（Execution Graph）
展示的是：一次对话从开始 → Agent 思考 → 调用 Tool → 结束的完整流程

二、图中每个节点是什么意思（从上到下）

`start`（起点）

每一次 New Thread / 新对话 都从这里开始
对应：你在右侧输入框点击 Interact

不可修改，是 LangGraph 内置节点

`agent`（核心节点）⭐

** Agent 本体**
实际执行的是：
- Gemini（或你配置的 LLM）
- - prompt
- - 当前 messages
- - memory（如果启用）

在代码中对应的就是：

graph = create_agent(...)

90% 的逻辑都发生在这里

`tools`（工具节点）

当 Agent 判断：

“我需要调用工具”
就会从 agent → tools
执行你定义的 Python 函数，例如：

def get_weather(city: str) -> str:
    ...

执行完后：

结果会返回给 agent
继续下一步判断

`agent ↔ tools` 的来回箭头（重点）

看到的是：

agent → tools → agent

这表示：

ReAct 模式
Agent 可以：
- 想一想
- 调用 Tool
- 再想一想
- 再决定是否继续

这是“智能行为”的来源

`end`（终点）

当 Agent 判断：

“我已经可以回答用户了”
流程走到这里
对话结束，结果返回 UI

三、图上的虚线 / 实线代表什么？

实线箭头：必经路径
虚线箭头：条件路径（Agent 决定）

例如：

不需要工具 → agent → __end__
需要工具 → agent → tools → agent

四、顶部几个按钮该怎么理解？

`Graph` / `Chat`

Graph：看流程结构（你现在看到的）
Chat：像 ChatGPT 一样直接聊天

`Memory`

是否启用 跨轮对话记忆
当前这张图是 无长期记忆 / in-memory

`Interrupts`

人工中断点
常用于：
- 人工审批
- 半自动 Agent

五、右侧 `New Thread` 是什么？

一个 Thread = 一次完整对话
每个 Thread：
- 有独立的 state
- 独立的执行路径
切换 Thread 就像切换聊天窗口

六、一句话理解

这张图 = 一个 LLM 驱动的状态机

Node = 方法

Edge = if / while

Tool = 外部服务

Thread = 会话上下文

七、测试验证agent

be401391c3f9624e628cbf.png)

12.langSmith 查看agent是否跑通

在这里插入图片描述

确认

看 Graph 是否走了：
```
agent → tools → agent → end
```
Tracing 看 Gemini 是否真的调用了 get_weather

这一张 Trace 说明 LangGraph + Gemini + Tool 调用已经 100% 跑通。
按 “从上到下、从左到右” 解释这一条执行链。

一、一句话结论

Gemini 成功理解了问题
成功决定 调用 get_weather 工具
工具被真实执行
结果回传给 Agent
Agent 生成最终自然语言回答

这是一个“标准、正确、健康”的 ReAct Agent Trace。

二、左侧 Waterfall（执行时间线）

这是执行顺序 + 耗时，从上到下就是时间顺序。

1️⃣ agent（第一次）

agent  0.97s
└─ call_model (gemini-2.5-flash)
└─ should_continue

含义：

Agent 读取用户输入
Gemini 推理：“我需要调用工具”
should_continue = True

👉 这一步 没有直接回答用户

2️⃣ tools → get_weather

tools
└─ get_weather

在右侧看到：

AI → get_weather
city: 广州

这说明：

Gemini 正确识别了参数
自动生成了 Tool Call
参数结构完全正确（city = 广州）

👉
这是 Tool Calling 成功的关键证据

3️⃣ TOOL 返回结果

城市 in 广州，今天天气晴朗!

这是 真实执行 Python 函数的返回值：

def get_weather(city: str) -> str:
    return f"城市 in {city}，今天天气晴朗!"

4️⃣ agent（第二次）

agent  0.97s
└─ call_model
└─ should_continue = False

含义：

Agent 拿到 tool 的返回结果
Gemini 决定：可以结束了
生成最终用户可读回答

三、右侧 Output（消息流，最重要）

消息顺序（非常标准）

1️⃣ HUMAN

今天广州的天气是什么

2️⃣ AI（Tool Call 意图）

get_weather(city=广州)

3️⃣ TOOL（真实执行结果）

城市 in 广州，今天天气晴朗!

4️⃣ AI（最终回答）

今天广州的天气晴朗!

👉 这正是 ReAct = Think → Act → Observe → Answer

四、为什么这条 Trace 非常“干净”

几个关键指标：

✅ 没有 retry
✅ 没有 hallucination
✅ 参数抽取正确
✅ 一次 tool call 就结束
✅ 最终回答简洁

这说明：

prompt 合理
tool 定义清晰
Gemini 2.5 Flash 表现稳定

客观说，
现在已经完成了：

✅ 最新 LangGraph
✅ Gemini 2.5 Flash
✅ ReAct Agent
✅ Tool Calling
✅ LangSmith Trace 阅读
✅ Graph / State / Tool 的完整闭环

Trace 就是 Agent 的“调用栈 + 日志”
你现在已经能读懂、判断、验证它是否“正确工作”。

2048 AI社区

有“AI”的1024 = 2048，欢迎大家加入2048 AI社区

更多推荐

又一家大厂宣布禁止用cursor！

2048 AI社区

JavaScript性能优化：让网页飞起来

通过解析接口契约（如OpenAPI规范），AI能够生成客户端SDK或Mock服务，使前后端开发人员能够并行工作，减少联调阶段的等待时间。生成的代码通常遵循统一的设计模式和命名规范，降低团队协作中的理解成本。在CI/CD管道中，AI可以分析测试覆盖率报告，自动补充边界条件测试用例，确保CRUD操作的健壮性。部分系统还能根据性能测试结果自动优化生成的SQL查询，例如添加缺失的索引提示，或重构JPA实体

2048 AI社区

【AI测试全栈：质量模型】5、全栈AI测试实战指南：电商推荐系统测试金字塔从理论到落地

《全栈AI测试实战指南》深入剖析电商推荐系统测试方法论，构建了从单元测试到社会测试的完整金字塔框架。文章首先解析推荐系统三层架构（召回-排序-后处理）及技术栈（Spring Boot+Python+Vue），针对AI模型不确定性、多端协同等挑战提出分层测试策略。重点展示了Python模型单元测试实践，涵盖前向传播、梯度验证等核心场景，提供50+可运行代码示例。通过Allure报告和真实业务用例，为

2048 AI社区

所有评论(0)

查看更多评论

Biehmltym

@qq_53433382

已为社区贡献5条内容

【AI】04AI Aent:十分钟跑通LangGraph项目:调用llm+agent开发+langSmith使用

Biehmltym

目录标题

1.conda python虚拟环境创建

2.安装 langgraph-cli

3.创建langgraph-cli应用

总结 CLI 实际完成的事情是：

5.用 PyCharm 打开创建的项目

6.langgraph项目解析

一、整体结论（“导航结论”）

二、项目结构总览（现在看到的）

三、最应该看的 5 个文件（按顺序）

① README.md —— 运行方式 & 官方意图

② .env.example —— LLM 配置模板（非常关键）

③ pyproject.toml —— 环境的配置文件:安装项目环境依赖 & Python 版本

④ langgraph.json —— LangGraph 项目的配置文件

⑤ ⭐ src/agent/graph.py —— 真正的“业务逻辑”

四、用 Java / 后端工程的角度理解（很重要）

7. pip install -e . 读取 pyproject.toml在虚拟环境安装依赖

8. .env配置环境文件：API KEY安装 llm对应的 LangChain Provider

test_gemini.py 项目目录下，可以新预先看llm能否跑通

9.graph.py：AI Agent智能体的业务代码

10.langgraph dev启动 LangGraph 开发服务器

二、默认行为（你不用额外配置）

三、常用参数（记这几个就够）

四、之后每次项目启动的正确使用顺序（推荐）

11.langSmith 使用

下面按“现在正在用 LangGraph Studio 的真实视角”，把这张图一次性讲清楚

一、这张图是什么？（一句话）

二、图中每个节点是什么意思（从上到下）

__start__（起点）

agent（核心节点）⭐

tools（工具节点）

agent ↔ tools 的来回箭头（重点）

__end__（终点）

三、图上的虚线 / 实线代表什么？

四、顶部几个按钮该怎么理解？

Graph / Chat

Memory

Interrupts

五、右侧 New Thread 是什么？

六、一句话理解

七、测试验证agent

12.langSmith 查看agent是否跑通

一、一句话结论

二、左侧 Waterfall（执行时间线）

1️⃣ agent（第一次）

2️⃣ tools → get_weather

3️⃣ TOOL 返回结果

4️⃣ agent（第二次）

三、右侧 Output（消息流，最重要）

消息顺序（非常标准）

四、为什么这条 Trace 非常“干净”

所有评论(0)

Biehmltym

① `README.md` —— 运行方式 & 官方意图

② `.env.example` —— LLM 配置模板（非常关键）

③ `pyproject.toml` —— 环境的配置文件:安装项目环境依赖 & Python 版本

④ `langgraph.json` —— LangGraph 项目的配置文件

⑤ ⭐ `src/agent/graph.py` —— 真正的“业务逻辑”

`start`（起点）

`agent`（核心节点）⭐

`tools`（工具节点）

`agent ↔ tools` 的来回箭头（重点）

`end`（终点）

`Graph` / `Chat`

`Memory`

`Interrupts`

五、右侧 `New Thread` 是什么？