这部分文档涵盖了带有工具包的代理——例如，应用于特定用例的代理。
有关代理工具包的完整列表，请参见下文
- Azure 认知服务工具包
- CSV代理
- Gmail 工具包
- 吉拉
- JSON 代理
- OpenAPI 代理
- 自然语言 API
- 熊猫数据框代理
- PlayWright 浏览器工具包
- PowerBI 数据集代理
- 蟒蛇代理
- Spark 数据框代理
- Spark SQL 代理
- SQL 数据库代理
- 向量存储代理
Azure 认知服务工具包
该工具包用于与 Azure 认知服务 API 交互以实现一些多模式功能。
目前此工具包中*绑捆**了四种工具：
- AzureCogsImageAnalysisTool：用于从图像中提取标题、对象、标签和文本。（注意：由于依赖azure-ai-vision包，目前仅在 Windows 和 Linux 上支持此工具，因此在 Mac OS 上尚不可用。）
- AzureCogsFormRecognizerTool：用于从文档中提取文本、表格和键值对。
- AzureCogsSpeech2TextTool：用于将语音转录为文本。
- AzureCogsText2SpeechTool：用于将文本合成为语音。
首先，您需要设置一个 Azure 帐户并创建一个认知服务资源。您可以按照此处的说明创建资源。
然后，您需要获取资源的端点、密钥和区域，并将它们设置为环境变量。您可以在资源的“密钥和端点”页面中找到它们。
# !pip install --upgrade azure-ai-formrecognizer > /dev/null
# !pip install --upgrade azure-cognitiveservices-speech > /dev/null
# For Windows/Linux
# !pip install --upgrade azure-ai-vision > /dev/null
import os
os.environ["OPENAI_API_KEY"] = "sk-"
os.environ["AZURE_COGS_KEY"] = ""
os.environ["AZURE_COGS_ENDPOINT"] = ""
os.environ["AZURE_COGS_REGION"] = ""
创建工具包
from langchain.agents.agent_toolkits import AzureCognitiveServicesToolkit
toolkit = AzureCognitiveServicesToolkit()
[tool.name for tool in toolkit.get_tools()]
['Azure Cognitive Services Image Analysis',
'Azure Cognitive Services Form Recognizer',
'Azure Cognitive Services Speech2Text',
'Azure Cognitive Services Text2Speech']
在代理中使用
from langchain import OpenAI
from langchain.agents import initialize_agent, AgentType
llm = OpenAI(temperature=0)
agent = initialize_agent(
tools=toolkit.get_tools(),
llm=llm,
agent=AgentType.STRUCTURED_CHAT_ZERO_SHOT_REACT_DESCRIPTION,
verbose=True,
)
agent.run("What can I make with these ingredients?"
"https://images.openai.com/blob/9ad5a2ab-041f-475f-ad6a-b51899c50182/ingredients.png")
> Entering new AgentExecutor chain...
Action:
```
{
"action": "Azure Cognitive Services Image Analysis",
"action_input": "https://images.openai.com/blob/9ad5a2ab-041f-475f-ad6a-b51899c50182/ingredients.png"
}
```
Observation: Caption: a group of eggs and flour in bowls
Objects: Egg, Egg, Food
Tags: dairy, ingredient, indoor, thickening agent, food, mixing bowl, powder, flour, egg, bowl
Thought: I can use the objects and tags to suggest recipes
Action:
```
{
"action": "Final Answer",
"action_input": "You can make pancakes, omelettes, or quiches with these ingredients!"
}
```
> Finished chain.
'You can make pancakes, omelettes, or quiches with these ingredients!'
audio_file = agent.run("Tell me a joke and read it out for me.")
> Entering new AgentExecutor chain...
Action:
```
{
"action": "Azure Cognitive Services Text2Speech",
"action_input": "Why did the chicken cross the playground? To get to the other slide!"
}
```
Observation: /tmp/tmpa3uu_j6b.wav
Thought: I have the audio file of the joke
Action:
```
{
"action": "Final Answer",
"action_input": "/tmp/tmpa3uu_j6b.wav"
}
```
> Finished chain.
'/tmp/tmpa3uu_j6b.wav'
from IPython import display
audio = display.Audio(audio_file)
display.display(audio)
CSV 代理
此笔记本展示了如何使用代理与 csv 交互。它主要针对问答进行了优化。
注意：此代理调用引擎盖下的 Pandas DataFrame 代理，后者又调用 Python 代理，后者执行 LLM 生成的 Python 代码——如果 LLM 生成的 Python 代码有害，这可能很糟糕。谨慎使用。
from langchain.agents import create_csv_agent
from langchain.llms import OpenAI
agent = create_csv_agent(OpenAI(temperature=0), 'titanic.csv', verbose=True)
agent.run("how many rows are there?")
> Entering new AgentExecutor chain...
Thought: I need to count the number of rows
Action: python_repl_ast
Action Input: df.shape[0]
Observation: 891
Thought: I now know the final answer
Final Answer: There are 891 rows.
> Finished chain.
'There are 891 rows.'
agent.run("how many people have more than 3 siblings")
> Entering new AgentExecutor chain...
Thought: I need to count the number of people with more than 3 siblings
Action: python_repl_ast
Action Input: df[df['SibSp'] > 3].shape[0]
Observation: 30
Thought: I now know the final answer
Final Answer: 30 people have more than 3 siblings.
> Finished chain.
'30 people have more than 3 siblings.'
agent.run("whats the square root of the average age?")
> Entering new AgentExecutor chain...
Thought: I need to calculate the average age first
Action: python_repl_ast
Action Input: df['Age'].mean()
Observation: 29.69911764705882
Thought: I now need to calculate the square root of the average age
Action: python_repl_ast
Action Input: math.sqrt(df['Age'].mean())
Observation: NameError("name 'math' is not defined")
Thought: I need to import the math library
Action: python_repl_ast
Action Input: import math
Observation:
Thought: I now need to calculate the square root of the average age
Action: python_repl_ast
Action Input: math.sqrt(df['Age'].mean())
Observation: 5.449689683556195
Thought: I now know the final answer
Final Answer: 5.449689683556195
> Finished chain.
'5.449689683556195'
多 CSV 示例
下一部分展示了代理如何与作为列表传入的多个 csv 文件进行交互。
agent = create_csv_agent(OpenAI(temperature=0), ['titanic.csv', 'titanic_age_fillna.csv'], verbose=True)
agent.run("how many rows in the age column are different?")
> Entering new AgentExecutor chain...
Thought: I need to compare the age columns in both dataframes
Action: python_repl_ast
Action Input: len(df1[df1['Age'] != df2['Age']])
Observation: 177
Thought: I now know the final answer
Final Answer: 177 rows in the age column are different.
> Finished chain.
'177 rows in the age column are different.'
Gmail 工具包
本笔记本介绍了将 LangChain 电子邮件连接到 Gmail API 的过程。
要使用此工具包，您需要设置Gmail API 文档中说明的凭据。*载下**credentials.json文件后，您就可以开始使用 Gmail API。完成后，我们将安装所需的库。
!pip install --upgrade google-api-python-client > /dev/null
!pip install --upgrade google-auth-oauthlib > /dev/null
!pip install --upgrade google-auth-httplib2 > /dev/null
!pip install beautifulsoup4 > /dev/null # This is optional but is useful for parsing HTML messages
创建工具包
默认情况下，工具包读取本地credentials.json文件。您也可以手动提供一个Credentials对象。
from langchain.agents.agent_toolkits import GmailToolkit
toolkit = GmailToolkit()
自定义身份验证
在幕后，googleapi资源是使用以下方法创建的。您可以手动构建googleapi资源以获得更多身份验证控制。
from langchain.tools.gmail.utils import build_resource_service, get_gmail_credentials
# Can review scopes here https://developers.google.com/gmail/api/auth/scopes
# For instance, readonly scope is 'https://www.googleapis.com/auth/gmail.readonly'
credentials = get_gmail_credentials(
token_file='token.json',
scopes=["https://mail.google.com/"],
client_secrets_file="credentials.json",
)
api_resource = build_resource_service(credentials=credentials)
toolkit = GmailToolkit(api_resource=api_resource)
tools = toolkit.get_tools()
tools
[GmailCreateDraft(name='create_gmail_draft', description='Use this tool to create a draft email with the provided message fields.', args_schema=<class 'langchain.tools.gmail.create_draft.CreateDraftSchema'>, return_direct=False, verbose=False, callbacks=None, callback_manager=None, api_resource=<googleapiclient.discovery.Resource object at 0x10e5c6d10>),
GmailSendMessage(name='send_gmail_message', description='Use this tool to send email messages. The input is the message, recipents', args_schema=None, return_direct=False, verbose=False, callbacks=None, callback_manager=None, api_resource=<googleapiclient.discovery.Resource object at 0x10e5c6d10>),
GmailSearch(name='search_gmail', description=('Use this tool to search for email messages or threads. The input must be a valid Gmail query. The output is a JSON list of the requested resource.',), args_schema=<class 'langchain.tools.gmail.search.SearchArgsSchema'>, return_direct=False, verbose=False, callbacks=None, callback_manager=None, api_resource=<googleapiclient.discovery.Resource object at 0x10e5c6d10>),
GmailGetMessage(name='get_gmail_message', description='Use this tool to fetch an email by message ID. Returns the thread ID, snipet, body, subject, and sender.', args_schema=<class 'langchain.tools.gmail.get_message.SearchArgsSchema'>, return_direct=False, verbose=False, callbacks=None, callback_manager=None, api_resource=<googleapiclient.discovery.Resource object at 0x10e5c6d10>),
GmailGetThread(name='get_gmail_thread', description=('Use this tool to search for email messages. The input must be a valid Gmail query. The output is a JSON list of messages.',), args_schema=<class 'langchain.tools.gmail.get_thread.GetThreadSchema'>, return_direct=False, verbose=False, callbacks=None, callback_manager=None, api_resource=<googleapiclient.discovery.Resource object at 0x10e5c6d10>)]
在代理中使用 from langchain import OpenAI from langchain.agents import initialize_agent, AgentType llm = OpenAI(temperature = 0 ) agent = initialize_agent( tools = toolkit . get_tools(), llm = llm, agent = AgentType . STRUCTURED_CHAT_ZERO_SHOT_REACT_DESCRIPTION, ) agent . run( "Create a gmail draft for me to edit of a letter from the perspective of a sentient parrot" " who is looking to collaborate on some research with her" " estranged friend, a cat. Under no circumstances may you send the message, however." ) WARNING:root:Failed to load default session, using empty session: 0 WARNING:root:Failed to persist run: {"detail":"Not Found"} 'I have created a draft email for you to edit. The draft Id is r5681294731961864018.' agent . run( "Could you search in my drafts for the latest email?" ) WARNING:root:Failed to load default session, using empty session: 0 WARNING:root:Failed to persist run: {"detail":"Not Found"} "The latest email in your drafts is from hopefulparrot@gmail.com with the subject 'Collaboration Opportunity'. The body of the email reads: 'Dear [Friend], I hope this letter finds you well. I am writing to you in the hopes of rekindling our friendship and to discuss the possibility of collaborating on some research together. I know that we have had our differences in the past, but I believe that we can put them aside and work together for the greater good. I look forward to hearing from you. Sincerely, [Parrot]'"
OpenAPI 代理
我们可以构造代理来使用任意 API，这里的 API 符合 OpenAPI/Swagger 规范。
第一个例子：分层规划代理
在此示例中，我们将考虑一种称为分层规划的方法，该方法在机器人技术中很常见，并且出现在 LLM X 机器人技术的近期作品中。我们将看到开始使用大量 API 规范并协助用户查询需要针对 API 执行多个步骤的可行方法。
这个想法很简单：为了在长序列行为上获得一致的代理行为并节省令牌，我们将分离关注点：“计划者”将负责调用哪些端点，“控制器”将负责如何调用他们。
在最初的实现中，规划器是一个 LLM 链，它具有上下文中每个端点的名称和简短描述。控制器是一个 LLM 代理，它仅使用特定计划的端点的文档进行实例化。要让这个工作非常稳健，还有很多事情要做:)
首先，让我们收集一些 OpenAPI 规范。
import os, yaml
!wget https://raw.githubusercontent.com/openai/openai-openapi/master/openapi.yaml
!mv openapi.yaml openai_openapi.yaml
!wget https://www.klarna.com/us/shopping/public/openai/v0/api-docs
!mv api-docs klarna_openapi.yaml
!wget https://raw.githubusercontent.com/APIs-guru/openapi-directory/main/APIs/spotify.com/1.0.0/openapi.yaml
!mv openapi.yaml spotify_openapi.yaml
--2026-03-18T10:41:54+00:00-- https://raw.githubusercontent.com/openai/openai-openapi/master/openapi.yaml
Resolving raw.githubusercontent.com (raw.githubusercontent.com)... 185.199.110.133, 185.199.109.133, 185.199.111.133, ...
Connecting to raw.githubusercontent.com (raw.githubusercontent.com)|185.199.110.133|:443... connected.
HTTP request sent, awaiting response... 200 OK
Length: 122995 (120K) [text/plain]
Saving to: ‘openapi.yaml’
openapi.yaml 100%[===================>] 120.11K --.-KB/s in 0.01s
2026-03-18T10:41:54+00:00 (10.4 MB/s) - ‘openapi.yaml’ saved [122995/122995]
--2026-03-18T10:41:54+00:00-- https://www.klarna.com/us/shopping/public/openai/v0/api-docs
Resolving www.klarna.com (www.klarna.com)... 52.84.150.34, 52.84.150.46, 52.84.150.61, ...
Connecting to www.klarna.com (www.klarna.com)|52.84.150.34|:443... connected.
HTTP request sent, awaiting response... 200 OK
Length: unspecified [application/json]
Saving to: ‘api-docs’
api-docs [ <=> ] 1.87K --.-KB/s in 0s
2026-03-18T10:41:54+00:00 (261 MB/s) - ‘api-docs’ saved [1916]
--2026-03-18T10:41:54+00:00-- https://raw.githubusercontent.com/APIs-guru/openapi-directory/main/APIs/spotify.com/1.0.0/openapi.yaml
Resolving raw.githubusercontent.com (raw.githubusercontent.com)... 185.199.110.133, 185.199.109.133, 185.199.111.133, ...
Connecting to raw.githubusercontent.com (raw.githubusercontent.com)|185.199.110.133|:443... connected.
HTTP request sent, awaiting response... 200 OK
Length: 286747 (280K) [text/plain]
Saving to: ‘openapi.yaml’
openapi.yaml 100%[===================>] 280.03K --.-KB/s in 0.02s
2026-03-18T10:41:54+00:00 (13.3 MB/s) - ‘openapi.yaml’ saved [286747/286747]
from langchain.agents.agent_toolkits.openapi.spec import reduce_openapi_spec
with open("openai_openapi.yaml") as f:
raw_openai_api_spec = yaml.load(f, Loader=yaml.Loader)
openai_api_spec = reduce_openapi_spec(raw_openai_api_spec)
with open("klarna_openapi.yaml") as f:
raw_klarna_api_spec = yaml.load(f, Loader=yaml.Loader)
klarna_api_spec = reduce_openapi_spec(raw_klarna_api_spec)
with open("spotify_openapi.yaml") as f:
raw_spotify_api_spec = yaml.load(f, Loader=yaml.Loader)
spotify_api_spec = reduce_openapi_spec(raw_spotify_api_spec)
我们将使用 Spotify API 作为有点复杂的 API 的示例之一。如果您想复制它，需要进行一些与身份验证相关的设置。
- 您必须在 Spotify 开发人员控制台中设置一个应用程序（记录在此处）以获取凭据：CLIENT_ID、CLIENT_SECRET和REDIRECT_URI。
- 要获取访问令牌（并使它们保持最新），您可以实施 oauth 流程，或者您可以使用spotipy. 如果您已将 Spotify 凭据设置为环境变量SPOTIPY_CLIENT_ID、SPOTIPY_CLIENT_SECRET和SPOTIPY_REDIRECT_URI，则可以使用以下辅助函数：
import spotipy.util as util
from langchain.requests import RequestsWrapper
def construct_spotify_auth_headers(raw_spec: dict):
scopes = list(raw_spec['components']['securitySchemes']['oauth_2_0']['flows']['authorizationCode']['scopes'].keys())
access_token = util.prompt_for_user_token(scope=','.join(scopes))
return {
'Authorization': f'Bearer {access_token}'
}
# Get API credentials.
headers = construct_spotify_auth_headers(raw_spotify_api_spec)
requests_wrapper = RequestsWrapper(headers=headers)
这个规格有多大？ endpoints = [ (route, operation) for route, operations in raw_spotify_api_spec[ "paths" ] . items() for operation in operations if operation in [ "get" , "post" ] ] len (endpoints) 63 import tiktoken enc = tiktoken . encoding_for_model( 'text-davinci-003' ) def count_tokens (s): return len (enc . encode(s)) count_tokens(yaml . dump(raw_spotify_api_spec)) 80326
让我们看一些例子！
从 GPT-4 开始。（GPT-3 系列正在进行一些稳健性迭代。）
from langchain.llms.openai import OpenAI
from langchain.agents.agent_toolkits.openapi import planner
llm = OpenAI(model_name="gpt-4", temperature=0.0)
/Users/jeremywelborn/src/langchain/langchain/llms/openai.py:169: UserWarning: You are trying to use a chat model. This way of initializing it is no longer supported. Instead, please use: `from langchain.chat_models import ChatOpenAI`
warnings.warn(
/Users/jeremywelborn/src/langchain/langchain/llms/openai.py:608: UserWarning: You are trying to use a chat model. This way of initializing it is no longer supported. Instead, please use: `from langchain.chat_models import ChatOpenAI`
warnings.warn(
spotify_agent = planner.create_openapi_agent(spotify_api_spec, requests_wrapper, llm)
user_query = "make me a playlist with the first song from kind of blue. call it machine blues."
spotify_agent.run(user_query)
> Entering new AgentExecutor chain...
Action: api_planner
Action Input: I need to find the right API calls to create a playlist with the first song from Kind of Blue and name it Machine Blues
Observation: 1. GET /search to search for the album "Kind of Blue"
2. GET /albums/{id}/tracks to get the tracks from the "Kind of Blue" album
3. GET /me to get the current user's information
4. POST /users/{user_id}/playlists to create a new playlist named "Machine Blues" for the current user
5. POST /playlists/{playlist_id}/tracks to add the first song from "Kind of Blue" to the "Machine Blues" playlist
Thought:I have the plan, now I need to execute the API calls.
Action: api_controller
Action Input: 1. GET /search to search for the album "Kind of Blue"
2. GET /albums/{id}/tracks to get the tracks from the "Kind of Blue" album
3. GET /me to get the current user's information
4. POST /users/{user_id}/playlists to create a new playlist named "Machine Blues" for the current user
5. POST /playlists/{playlist_id}/tracks to add the first song from "Kind of Blue" to the "Machine Blues" playlist
> Entering new AgentExecutor chain...
Action: requests_get
Action Input: {"url": "https://api.spotify.com/v1/search?q=Kind%20of%20Blue&type=album", "output_instructions": "Extract the id of the first album in the search results"}
Observation: 1weenld61qoidwYuZ1GESA
Thought:Action: requests_get
Action Input: {"url": "https://api.spotify.com/v1/albums/1weenld61qoidwYuZ1GESA/tracks", "output_instructions": "Extract the id of the first track in the album"}
Observation: 7q3kkfAVpmcZ8g6JUThi3o
Thought:Action: requests_get
Action Input: {"url": "https://api.spotify.com/v1/me", "output_instructions": "Extract the id of the current user"}
Observation: 22rhrz4m4kvpxlsb5hezokzwi
Thought:Action: requests_post
Action Input: {"url": "https://api.spotify.com/v1/users/22rhrz4m4kvpxlsb5hezokzwi/playlists", "data": {"name": "Machine Blues"}, "output_instructions": "Extract the id of the created playlist"}
Observation: 7lzoEi44WOISnFYlrAIqyX
Thought:Action: requests_post
Action Input: {"url": "https://api.spotify.com/v1/playlists/7lzoEi44WOISnFYlrAIqyX/tracks", "data": {"uris": ["spotify:track:7q3kkfAVpmcZ8g6JUThi3o"]}, "output_instructions": "Confirm that the track was added to the playlist"}
Observation: The track was added to the playlist, confirmed by the snapshot_id: MiwxODMxNTMxZTFlNzg3ZWFlZmMxYTlmYWQyMDFiYzUwNDEwMTAwZmE1.
Thought:I am finished executing the plan.
Final Answer: The first song from the "Kind of Blue" album has been added to the "Machine Blues" playlist.
> Finished chain.
Observation: The first song from the "Kind of Blue" album has been added to the "Machine Blues" playlist.
Thought:I am finished executing the plan and have created the playlist with the first song from Kind of Blue.
Final Answer: I have created a playlist called "Machine Blues" with the first song from the "Kind of Blue" album.
> Finished chain.
'I have created a playlist called "Machine Blues" with the first song from the "Kind of Blue" album.'
user_query = "give me a song I'd like, make it blues-ey"
spotify_agent.run(user_query)
> Entering new AgentExecutor chain...
Action: api_planner
Action Input: I need to find the right API calls to get a blues song recommendation for the user
Observation: 1. GET /me to get the current user's information
2. GET /recommendations/available-genre-seeds to retrieve a list of available genres
3. GET /recommendations with the seed_genre parameter set to "blues" to get a blues song recommendation for the user
Thought:I have the plan, now I need to execute the API calls.
Action: api_controller
Action Input: 1. GET /me to get the current user's information
2. GET /recommendations/available-genre-seeds to retrieve a list of available genres
3. GET /recommendations with the seed_genre parameter set to "blues" to get a blues song recommendation for the user
> Entering new AgentExecutor chain...
Action: requests_get
Action Input: {"url": "https://api.spotify.com/v1/me", "output_instructions": "Extract the user's id and username"}
Observation: ID: 22rhrz4m4kvpxlsb5hezokzwi, Username: Jeremy Welborn
Thought:Action: requests_get
Action Input: {"url": "https://api.spotify.com/v1/recommendations/available-genre-seeds", "output_instructions": "Extract the list of available genres"}
Observation: acoustic, afrobeat, alt-rock, alternative, ambient, anime, black-metal, bluegrass, blues, bossanova, brazil, breakbeat, british, cantopop, chicago-house, children, chill, classical, club, comedy, country, dance, dancehall, death-metal, deep-house, detroit-techno, disco, disney, drum-and-bass, dub, dubstep, edm, electro, electronic, emo, folk, forro, french, funk, garage, german, gospel, goth, grindcore, groove, grunge, guitar, happy, hard-rock, hardcore, hardstyle, heavy-metal, hip-hop, holidays, honky-tonk, house, idm, indian, indie, indie-pop, industrial, iranian, j-dance, j-idol, j-pop, j-rock, jazz, k-pop, kids, latin, latino, malay, mandopop, metal, metal-misc, metalcore, minimal-techno, movies, mpb, new-age, new-release, opera, pagode, party, philippines-
Thought:
Retrying langchain.llms.openai.completion_with_retry.<locals>._completion_with_retry in 4.0 seconds as it raised RateLimitError: That model is currently overloaded with other requests. You can retry your request, or contact us through our help center at help.openai.com if the error persists. (Please include the request ID 2167437a0072228238f3c0c5b3882764 in your message.).
Action: requests_get
Action Input: {"url": "https://api.spotify.com/v1/recommendations?seed_genres=blues", "output_instructions": "Extract the list of recommended tracks with their ids and names"}
Observation: [
{
id: '03lXHmokj9qsXspNsPoirR',
name: 'Get Away Jordan'
}
]
Thought:I am finished executing the plan.
Final Answer: The recommended blues song for user Jeremy Welborn (ID: 22rhrz4m4kvpxlsb5hezokzwi) is "Get Away Jordan" with the track ID: 03lXHmokj9qsXspNsPoirR.
> Finished chain.
Observation: The recommended blues song for user Jeremy Welborn (ID: 22rhrz4m4kvpxlsb5hezokzwi) is "Get Away Jordan" with the track ID: 03lXHmokj9qsXspNsPoirR.
Thought:I am finished executing the plan and have the information the user asked for.
Final Answer: The recommended blues song for you is "Get Away Jordan" with the track ID: 03lXHmokj9qsXspNsPoirR.
> Finished chain.
'The recommended blues song for you is "Get Away Jordan" with the track ID: 03lXHmokj9qsXspNsPoirR.'
尝试另一个 API。
headers = {
"Authorization": f"Bearer {os.getenv('OPENAI_API_KEY')}"
}
openai_requests_wrapper=RequestsWrapper(headers=headers)
# Meta!
llm = OpenAI(model_name="gpt-4", temperature=0.25)
openai_agent = planner.create_openapi_agent(openai_api_spec, openai_requests_wrapper, llm)
user_query = "generate a short piece of advice"
openai_agent.run(user_query)
> Entering new AgentExecutor chain...
Action: api_planner
Action Input: I need to find the right API calls to generate a short piece of advice
Observation: 1. GET /engines to retrieve the list of available engines
2. POST /completions with the selected engine and a prompt for generating a short piece of advice
Thought:I have the plan, now I need to execute the API calls.
Action: api_controller
Action Input: 1. GET /engines to retrieve the list of available engines
2. POST /completions with the selected engine and a prompt for generating a short piece of advice
> Entering new AgentExecutor chain...
Action: requests_get
Action Input: {"url": "https://api.openai.com/v1/engines", "output_instructions": "Extract the ids of the engines"}
Observation: babbage, davinci, text-davinci-edit-001, babbage-code-search-code, text-similarity-babbage-001, code-davinci-edit-001, text-davinci-001, ada, babbage-code-search-text, babbage-similarity, whisper-1, code-search-babbage-text-001, text-curie-001, code-search-babbage-code-001, text-ada-001, text-embedding-ada-002, text-similarity-ada-001, curie-instruct-beta, ada-code-search-code, ada-similarity, text-davinci-003, code-search-ada-text-001, text-search-ada-query-001, davinci-search-document, ada-code-search-text, text-search-ada-doc-001, davinci-instruct-beta, text-similarity-curie-001, code-search-ada-code-001
Thought:I will use the "davinci" engine to generate a short piece of advice.
Action: requests_post
Action Input: {"url": "https://api.openai.com/v1/completions", "data": {"engine": "davinci", "prompt": "Give me a short piece of advice on how to be more productive."}, "output_instructions": "Extract the text from the first choice"}
Observation: "you must provide a model parameter"
Thought:!! Could not _extract_tool_and_input from "I cannot finish executing the plan without knowing how to provide the model parameter correctly." in _get_next_action
I cannot finish executing the plan without knowing how to provide the model parameter correctly.
> Finished chain.
Observation: I need more information on how to provide the model parameter correctly in the POST request to generate a short piece of advice.
Thought:I need to adjust my plan to include the model parameter in the POST request.
Action: api_planner
Action Input: I need to find the right API calls to generate a short piece of advice, including the model parameter in the POST request
Observation: 1. GET /models to retrieve the list of available models
2. Choose a suitable model from the list
3. POST /completions with the chosen model as a parameter to generate a short piece of advice
Thought:I have an updated plan, now I need to execute the API calls.
Action: api_controller
Action Input: 1. GET /models to retrieve the list of available models
2. Choose a suitable model from the list
3. POST /completions with the chosen model as a parameter to generate a short piece of advice
> Entering new AgentExecutor chain...
Action: requests_get
Action Input: {"url": "https://api.openai.com/v1/models", "output_instructions": "Extract the ids of the available models"}
Observation: babbage, davinci, text-davinci-edit-001, babbage-code-search-code, text-similarity-babbage-001, code-davinci-edit-001, text-davinci-edit-001, ada
Thought:Action: requests_post
Action Input: {"url": "https://api.openai.com/v1/completions", "data": {"model": "davinci", "prompt": "Give me a short piece of advice on how to improve communication skills."}, "output_instructions": "Extract the text from the first choice"}
Observation: "I'd like to broaden my horizon.\n\nI was trying to"
Thought:I cannot finish executing the plan without knowing some other information.
Final Answer: The generated text is not a piece of advice on improving communication skills. I would need to retry the API call with a different prompt or model to get a more relevant response.
> Finished chain.
Observation: The generated text is not a piece of advice on improving communication skills. I would need to retry the API call with a different prompt or model to get a more relevant response.
Thought:I need to adjust my plan to include a more specific prompt for generating a short piece of advice on improving communication skills.
Action: api_planner
Action Input: I need to find the right API calls to generate a short piece of advice on improving communication skills, including the model parameter in the POST request
Observation: 1. GET /models to retrieve the list of available models
2. Choose a suitable model for generating text (e.g., text-davinci-002)
3. POST /completions with the chosen model and a prompt related to improving communication skills to generate a short piece of advice
Thought:I have an updated plan, now I need to execute the API calls.
Action: api_controller
Action Input: 1. GET /models to retrieve the list of available models
2. Choose a suitable model for generating text (e.g., text-davinci-002)
3. POST /completions with the chosen model and a prompt related to improving communication skills to generate a short piece of advice
> Entering new AgentExecutor chain...
Action: requests_get
Action Input: {"url": "https://api.openai.com/v1/models", "output_instructions": "Extract the names of the models"}
Observation: babbage, davinci, text-davinci-edit-001, babbage-code-search-code, text-similarity-babbage-001, code-davinci-edit-001, text-davinci-edit-001, ada
Thought:Action: requests_post
Action Input: {"url": "https://api.openai.com/v1/completions", "data": {"model": "text-davinci-002", "prompt": "Give a short piece of advice on how to improve communication skills"}, "output_instructions": "Extract the text from the first choice"}
Observation: "Some basic advice for improving communication skills would be to make sure to listen"
Thought:I am finished executing the plan.
Final Answer: Some basic advice for improving communication skills would be to make sure to listen.
> Finished chain.
Observation: Some basic advice for improving communication skills would be to make sure to listen.
Thought:I am finished executing the plan and have the information the user asked for.
Final Answer: A short piece of advice for improving communication skills is to make sure to listen.
> Finished chain.
'A short piece of advice for improving communication skills is to make sure to listen.'
需要一段时间才能到达那里！
第二个例子：“json explorer”代理
这是一个不是特别实用但很整洁的代理！代理可以访问 2 个工具包。一个包含与 json 交互的工具：一个工具用于列出 json 对象的键，另一个工具用于获取给定键的值。另一个工具包包含requests用于发送 GET 和 POST 请求的包装器。这个代理消耗了很多对语言模型的调用，但做得非常好。
from langchain.agents import create_openapi_agent
from langchain.agents.agent_toolkits import OpenAPIToolkit
from langchain.llms.openai import OpenAI
from langchain.requests import TextRequestsWrapper
from langchain.tools.json.tool import JsonSpec
with open("openai_openapi.yaml") as f:
data = yaml.load(f, Loader=yaml.FullLoader)
json_spec=JsonSpec(dict_=data, max_value_length=4000)
openapi_toolkit = OpenAPIToolkit.from_llm(OpenAI(temperature=0), json_spec, openai_requests_wrapper, verbose=True)
openapi_agent_executor = create_openapi_agent(
llm=OpenAI(temperature=0),
toolkit=openapi_toolkit,
verbose=True
)
openapi_agent_executor.run("Make a post request to openai /completions. The prompt should be 'tell me a joke.'")
> Entering new AgentExecutor chain...
Action: json_explorer
Action Input: What is the base url for the API?
> Entering new AgentExecutor chain...
Action: json_spec_list_keys
Action Input: data
Observation: ['openapi', 'info', 'servers', 'tags', 'paths', 'components', 'x-oaiMeta']
Thought: I should look at the servers key to see what the base url is
Action: json_spec_list_keys
Action Input: data["servers"][0]
Observation: ValueError('Value at path `data["servers"][0]` is not a dict, get the value directly.')
Thought: I should get the value of the servers key
Action: json_spec_get_value
Action Input: data["servers"][0]
Observation: {'url': 'https://api.openai.com/v1'}
Thought: I now know the base url for the API
Final Answer: The base url for the API is https://api.openai.com/v1
> Finished chain.
Observation: The base url for the API is https://api.openai.com/v1
Thought: I should find the path for the /completions endpoint.
Action: json_explorer
Action Input: What is the path for the /completions endpoint?
> Entering new AgentExecutor chain...
Action: json_spec_list_keys
Action Input: data
Observation: ['openapi', 'info', 'servers', 'tags', 'paths', 'components', 'x-oaiMeta']
Thought: I should look at the paths key to see what endpoints exist
Action: json_spec_list_keys
Action Input: data["paths"]
Observation: ['/engines', '/engines/{engine_id}', '/completions', '/chat/completions', '/edits', '/images/generations', '/images/edits', '/images/variations', '/embeddings', '/audio/transcriptions', '/audio/translations', '/engines/{engine_id}/search', '/files', '/files/{file_id}', '/files/{file_id}/content', '/answers', '/classifications', '/fine-tunes', '/fine-tunes/{fine_tune_id}', '/fine-tunes/{fine_tune_id}/cancel', '/fine-tunes/{fine_tune_id}/events', '/models', '/models/{model}', '/moderations']
Thought: I now know the path for the /completions endpoint
Final Answer: The path for the /completions endpoint is data["paths"][2]
> Finished chain.
Observation: The path for the /completions endpoint is data["paths"][2]
Thought: I should find the required parameters for the POST request.
Action: json_explorer
Action Input: What are the required parameters for a POST request to the /completions endpoint?
> Entering new AgentExecutor chain...
Action: json_spec_list_keys
Action Input: data
Observation: ['openapi', 'info', 'servers', 'tags', 'paths', 'components', 'x-oaiMeta']
Thought: I should look at the paths key to see what endpoints exist
Action: json_spec_list_keys
Action Input: data["paths"]
Observation: ['/engines', '/engines/{engine_id}', '/completions', '/chat/completions', '/edits', '/images/generations', '/images/edits', '/images/variations', '/embeddings', '/audio/transcriptions', '/audio/translations', '/engines/{engine_id}/search', '/files', '/files/{file_id}', '/files/{file_id}/content', '/answers', '/classifications', '/fine-tunes', '/fine-tunes/{fine_tune_id}', '/fine-tunes/{fine_tune_id}/cancel', '/fine-tunes/{fine_tune_id}/events', '/models', '/models/{model}', '/moderations']
Thought: I should look at the /completions endpoint to see what parameters are required
Action: json_spec_list_keys
Action Input: data["paths"]["/completions"]
Observation: ['post']
Thought: I should look at the post key to see what parameters are required
Action: json_spec_list_keys
Action Input: data["paths"]["/completions"]["post"]
Observation: ['operationId', 'tags', 'summary', 'requestBody', 'responses', 'x-oaiMeta']
Thought: I should look at the requestBody key to see what parameters are required
Action: json_spec_list_keys
Action Input: data["paths"]["/completions"]["post"]["requestBody"]
Observation: ['required', 'content']
Thought: I should look at the content key to see what parameters are required
Action: json_spec_list_keys
Action Input: data["paths"]["/completions"]["post"]["requestBody"]["content"]
Observation: ['application/json']
Thought: I should look at the application/json key to see what parameters are required
Action: json_spec_list_keys
Action Input: data["paths"]["/completions"]["post"]["requestBody"]["content"]["application/json"]
Observation: ['schema']
Thought: I should look at the schema key to see what parameters are required
Action: json_spec_list_keys
Action Input: data["paths"]["/completions"]["post"]["requestBody"]["content"]["application/json"]["schema"]
Observation: ['$ref']
Thought: I should look at the $ref key to see what parameters are required
Action: json_spec_list_keys
Action Input: data["paths"]["/completions"]["post"]["requestBody"]["content"]["application/json"]["schema"]["$ref"]
Observation: ValueError('Value at path `data["paths"]["/completions"]["post"]["requestBody"]["content"]["application/json"]["schema"]["$ref"]` is not a dict, get the value directly.')
Thought: I should look at the $ref key to get the value directly
Action: json_spec_get_value
Action Input: data["paths"]["/completions"]["post"]["requestBody"]["content"]["application/json"]["schema"]["$ref"]
Observation: #/components/schemas/CreateCompletionRequest
Thought: I should look at the CreateCompletionRequest schema to see what parameters are required
Action: json_spec_list_keys
Action Input: data["components"]["schemas"]["CreateCompletionRequest"]
Observation: ['type', 'properties', 'required']
Thought: I should look at the required key to see what parameters are required
Action: json_spec_get_value
Action Input: data["components"]["schemas"]["CreateCompletionRequest"]["required"]
Observation: ['model']
Thought: I now know the final answer
Final Answer: The required parameters for a POST request to the /completions endpoint are 'model'.
> Finished chain.
Observation: The required parameters for a POST request to the /completions endpoint are 'model'.
Thought: I now know the parameters needed to make the request.
Action: requests_post
Action Input: { "url": "https://api.openai.com/v1/completions", "data": { "model": "davinci", "prompt": "tell me a joke" } }
Observation: {"id":"cmpl-70Ivzip3dazrIXU8DSVJGzFJj2rdv","object":"text_completion","created":1680307139,"model":"davinci","choices":[{"text":" with mummy not there”\n\nYou dig deep and come up with,","index":0,"logprobs":null,"finish_reason":"length"}],"usage":{"prompt_tokens":4,"completion_tokens":16,"total_tokens":20}}
Thought: I now know the final answer.
Final Answer: The response of the POST request is {"id":"cmpl-70Ivzip3dazrIXU8DSVJGzFJj2rdv","object":"text_completion","created":1680307139,"model":"davinci","choices":[{"text":" with mummy not there”\n\nYou dig deep and come up with,","index":0,"logprobs":null,"finish_reason":"length"}],"usage":{"prompt_tokens":4,"completion_tokens":16,"total_tokens":20}}
> Finished chain.
'The response of the POST request is {"id":"cmpl-70Ivzip3dazrIXU8DSVJGzFJj2rdv","object":"text_completion","created":1680307139,"model":"davinci","choices":[{"text":" with mummy not there”\\n\\nYou dig deep and come up with,","index":0,"logprobs":null,"finish_reason":"length"}],"usage":{"prompt_tokens":4,"completion_tokens":16,"total_tokens":20}}'
Vectorstore 代理
此笔记本展示了一种代理，旨在从一个或多个矢量存储中检索信息，无论是否有源。
创建矢量存储 from langchain.embeddings.openai import OpenAIEmbeddings from langchain.vectorstores import Chroma from langchain.text_splitter import CharacterTextSplitter from langchain import OpenAI, VectorDBQA llm = OpenAI(temperature = 0 ) from langchain.document_loaders import TextLoader loader = TextLoader( '../../../state_of_the_union.txt' ) documents = loader . load() text_splitter = CharacterTextSplitter(chunk_size = 1000 , chunk_overlap = 0 ) texts = text_splitter . split_documents(documents) embeddings = OpenAIEmbeddings() state_of_union_store = Chroma . from_documents(texts, embeddings, collection_name ="state-of-union" ) Running Chroma using direct local API. Using DuckDB in-memory for database. Data will be transient. from langchain.document_loaders import WebBaseLoader loader = WebBaseLoader( "https://beta.ruff.rs/docs/faq/" ) docs = loader . load() ruff_texts = text_splitter . split_documents(docs) ruff_store = Chroma . from_documents(ruff_texts, embeddings, collection_name ="ruff" ) Running Chroma using direct local API. Using DuckDB in-memory for database. Data will be transient.
初始化工具包和代理
首先，我们将创建一个具有单个 vectorstore 的代理。
from langchain.agents.agent_toolkits import (
create_vectorstore_agent,
VectorStoreToolkit,
VectorStoreInfo,
)
vectorstore_info = VectorStoreInfo(
name="state_of_union_address",
description="the most recent state of the Union adress",
vectorstore=state_of_union_store
)
toolkit = VectorStoreToolkit(vectorstore_info=vectorstore_info)
agent_executor = create_vectorstore_agent(
llm=llm,
toolkit=toolkit,
verbose=True
)
例子
agent_executor . run( "What did biden say about ketanji brown jackson is the state of the union address?" )
> Entering new AgentExecutor chain... I need to find the answer in the state of the union address Action: state_of_union_address Action Input: What did biden say about ketanji brown jackson Observation: Biden said that Ketanji Brown Jackson is one of the nation's top legal minds and that she will continue Justice Breyer's legacy of excellence. Thought: I now know the final answer Final Answer: Biden said that Ketanji Brown Jackson is one of the nation's top legal minds and that she will continue Justice Breyer's legacy of excellence. > Finished chain.
"Biden said that Ketanji Brown Jackson is one of the nation's top legal minds and that she will continue Justice Breyer's legacy of excellence."
agent_executor . run( "What did biden say about ketanji brown jackson is the state of the union address? List the source." )
> Entering new AgentExecutor chain... I need to use the state_of_union_address_with_sources tool to answer this question. Action: state_of_union_address_with_sources Action Input: What did biden say about ketanji brown jackson Observation: {"answer": " Biden said that he nominated Circuit Court of Appeals Judge Ketanji Brown Jackson to the United States Supreme Court, and that she is one of the nation's top legal minds who will continue Justice Breyer's legacy of excellence.\n", "sources": "../../state_of_the_union.txt"} Thought: I now know the final answer Final Answer: Biden said that he nominated Circuit Court of Appeals Judge Ketanji Brown Jackson to the United States Supreme Court, and that she is one of the nation's top legal minds who will continue Justice Breyer's legacy of excellence. Sources: ../../state_of_the_union.txt > Finished chain.
"Biden said that he nominated Circuit Court of Appeals Judge Ketanji Brown Jackson to the United States Supreme Court, and that she is one of the nation's top legal minds who will continue Justice Breyer's legacy of excellence. Sources: ../../state_of_the_union.txt"
多个向量存储
我们还可以轻松地使用它初始化具有多个矢量存储的代理，并使用代理在它们之间进行路由。去做这个。该代理针对路由进行了优化，因此它是一个不同的工具包和初始化程序。
from langchain.agents.agent_toolkits import (
create_vectorstore_router_agent,
VectorStoreRouterToolkit,
VectorStoreInfo,
)
ruff_vectorstore_info = VectorStoreInfo(
name="ruff",
description="Information about the Ruff python linting library",
vectorstore=ruff_store
)
router_toolkit = VectorStoreRouterToolkit(
vectorstores=[vectorstore_info, ruff_vectorstore_info],
llm=llm
)
agent_executor = create_vectorstore_router_agent(
llm=llm,
toolkit=router_toolkit,
verbose=True
)
例子
agent_executor.run("What did biden say about ketanji brown jackson is the state of the union address?")
> Entering new AgentExecutor chain...
I need to use the state_of_union_address tool to answer this question.
Action: state_of_union_address
Action Input: What did biden say about ketanji brown jackson
Observation: Biden said that Ketanji Brown Jackson is one of the nation's top legal minds and that she will continue Justice Breyer's legacy of excellence.
Thought: I now know the final answer
Final Answer: Biden said that Ketanji Brown Jackson is one of the nation's top legal minds and that she will continue Justice Breyer's legacy of excellence.
> Finished chain.
"Biden said that Ketanji Brown Jackson is one of the nation's top legal minds and that she will continue Justice Breyer's legacy of excellence."
agent_executor.run("What tool does ruff use to run over Jupyter Notebooks?")
> Entering new AgentExecutor chain...
I need to find out what tool ruff uses to run over Jupyter Notebooks
Action: ruff
Action Input: What tool does ruff use to run over Jupyter Notebooks?
Observation: Ruff is integrated into nbQA, a tool for running linters and code formatters over Jupyter Notebooks. After installing ruff and nbqa, you can run Ruff over a notebook like so: > nbqa ruff Untitled.ipynb
Thought: I now know the final answer
Final Answer: Ruff is integrated into nbQA, a tool for running linters and code formatters over Jupyter Notebooks. After installing ruff and nbqa, you can run Ruff over a notebook like so: > nbqa ruff Untitled.ipynb
> Finished chain.
'Ruff is integrated into nbQA, a tool for running linters and code formatters over Jupyter Notebooks. After installing ruff and nbqa, you can run Ruff over a notebook like so: > nbqa ruff Untitled.ipynb'
agent_executor.run("What tool does ruff use to run over Jupyter Notebooks? Did the president mention that tool in the state of the union?")
> Entering new AgentExecutor chain...
I need to find out what tool ruff uses and if the president mentioned it in the state of the union.
Action: ruff
Action Input: What tool does ruff use to run over Jupyter Notebooks?
Observation: Ruff is integrated into nbQA, a tool for running linters and code formatters over Jupyter Notebooks. After installing ruff and nbqa, you can run Ruff over a notebook like so: > nbqa ruff Untitled.ipynb
Thought: I need to find out if the president mentioned nbQA in the state of the union.
Action: state_of_union_address
Action Input: Did the president mention nbQA in the state of the union?
Observation: No, the president did not mention nbQA in the state of the union.
Thought: I now know the final answer.
Final Answer: No, the president did not mention nbQA in the state of the union.
> Finished chain.
'No, the president did not mention nbQA in the state of the union.'