integry 0.0.14

The official Python library for the Integry API

Integry Python API Library

The Python API library allows access to Integry REST API from Python programs.

Installation

# install from PyPI
pip install integry

Usage with Agent Frameworks

1. LangChain/LangGraph

import os
from integry import Integry
from langchain_core.messages import SystemMessage, HumanMessage
from langchain_openai import ChatOpenAI
from langchain_core.tools import StructuredTool
from langgraph.prebuilt import create_react_agent

user_id = "your user's ID"

# Initialize the client
integry = Integry(
    app_key=os.environ.get("INTEGRY_APP_KEY"),
    app_secret=os.environ.get("INTEGRY_APP_SECRET"),
)

slack_post_message = await integry.functions.get("slack-post-message", user_id)

llm = ChatOpenAI(
    model="gpt-4o",
    api_key=os.environ.get("OPENAI_API_KEY"),
)

tool = slack_post_message.get_langchain_tool(StructuredTool.from_function, user_id)

agent = create_react_agent(
    tools=[tool],
    model=llm,
)

await agent.ainvoke({
    "messages": [
        SystemMessage(content="You are a helpful assistant"),
        HumanMessage(content="Say hello to my team on slack"),
    ]
})

2. CrewAI

import os
from integry import Integry
from crewai import Agent, Task, Crew, LLM
from crewai.tools.structured_tool import CrewStructuredTool

user_id = "your user's ID"

# Initialize the client
integry = Integry(
    app_key=os.environ.get("INTEGRY_APP_KEY"),
    app_secret=os.environ.get("INTEGRY_APP_SECRET"),
)

slack_post_message = await integry.functions.get("slack-post-message", user_id)

tools = [
    slack_post_message.get_langchain_tool(CrewStructuredTool.from_function, user_id)
]

llm = LLM(
    model="gpt-4o",
    temperature=0,
    base_url="https://api.openai.com/v1",
    api_key=os.environ.get("OPENAI_API_KEY"),
)

crewai_agent = Agent(
    role="Integration Assistant",
    goal="Help users achieve their goal by performing their required task in various apps",
    backstory="You are a virtual assistant with access to various apps and services. You are known for your ability to connect to any app and perform any task.",
    verbose=True,
    tools=tools,
    llm=llm,
)

task = Task(
    description="Say hello to my team on slack",
    agent=crewai_agent,
    expected_output="Result of the task",
)

crew = Crew(agents=[crewai_agent], tasks=[task])

result = crew.kickoff()

3. AutoGen

import os
from integry import Integry
from autogen import ConversableAgent, register_function

user_id = "your user's ID"

# Initialize the client
integry = Integry(
    app_key=os.environ.get("INTEGRY_APP_KEY"),
    app_secret=os.environ.get("INTEGRY_APP_SECRET"),
)

function = await integry.functions.get("slack-post-message", user_id)

llm_config = {"config_list": [{"model": "gpt-4o", "api_key": os.environ.get("OPENAI_API_KEY")}]}

assistant = ConversableAgent(
    name="Assistant",
    system_message="You are a helpful integrations assistant. "
    "You can help users perform tasks in various apps. "
    "Return 'TERMINATE' when the task is done.",
    llm_config=llm_config,
)

user_proxy = ConversableAgent(
    name="User",
    llm_config=False,
    is_termination_msg=lambda msg: msg.get("content") is not None
    and "TERMINATE" in msg["content"],
    human_input_mode="NEVER",
)

function.register_with_autogen_agents(
    register_function,
    caller=assistant,
    executor=user_proxy,
    user_id=user_id,
)

chat_result = await user_proxy.a_initiate_chat(
    assistant,
    message="Say hello to my team on slack",
)

4. LlamaIndex

import os
from integry import Integry
from llama_index.core.tools import FunctionTool, ToolMetadata
from llama_index.llms.openai import OpenAI
from llama_index.core.agent import ReActAgent

user_id = "your user's ID"

# Initialize the client
integry = Integry(
    app_key=os.environ.get("INTEGRY_APP_KEY"),
    app_secret=os.environ.get("INTEGRY_APP_SECRET"),
)

slack_post_message = await integry.functions.get("slack-post-message", user_id)

llm = OpenAI(model="gpt-4o", temperature=0, api_key=os.environ.get("OPENAI_API_KEY"))

tools = [
    slack_post_message.get_llamaindex_tool(FunctionTool.from_defaults, ToolMetadata, user_id)
]

agent = ReActAgent.from_tools(tools=tools, llm=llm, verbose=True)

task = "Say hello to my team on slack."

result = await agent.achat(task)

5. Haystack

import os
from integry import Integry
from haystack.components.generators.chat import OpenAIChatGenerator
from haystack.dataclasses import ChatMessage
from haystack.components.tools import ToolInvoker
from haystack.tools import Tool

user_id = "your user's ID"

os.environ.get("OPENAI_API_KEY")

# Initialize the client
integry = Integry(
    app_key=os.environ.get("INTEGRY_APP_KEY"),
    app_secret=os.environ.get("INTEGRY_APP_SECRET"),
)


slack_post_message = await integry.functions.get("slack-post-message", user_id)

tool = slack_post_message.get_haystack_tool(Tool, user_id)

chat_generator = OpenAIChatGenerator(model="gpt-4o-mini", tools=[tool])

tool_invoker = ToolInvoker(tools=[tool])

user_message = ChatMessage.from_user("Say hello to my team on slack.")

replies = chat_generator.run(messages=[user_message])["replies"]

if replies[0].tool_calls:
    tool_messages = tool_invoker.run(messages=replies)["tool_messages"]
    print(f"tool messages: {tool_messages}")

6. Smolagents

import os
from smolagents import tool, CodeAgent, HfApiModel
from integry import Integry

user_id = "your user's ID"
hugging_face_token = os.environ.get("HUGGING_FACE_TOKEN")

# Initialize the client
integry = Integry(
    app_key=os.environ.get("INTEGRY_APP_KEY"),
    app_secret=os.environ.get("INTEGRY_APP_SECRET"),
)

slack_post_message = await integry.functions.get("slack-post-message", user_id)

slack_tool = slack_post_message.get_smolagent_tool(tool, user_id)

agent = CodeAgent(tools=[slack_tool], model=HfApiModel(token=hugging_face_token))

agent.run("Say hello to my team on slack.")

7. Litellm

import litellm
from integry import Integry
from integry import handle_litellm_tool_calls

user_id = "your user's ID"

os.environ.get("OPENAI_API_KEY")

# Initialize the client
integry = Integry(
    app_key=os.environ.get("INTEGRY_APP_KEY"),
    app_secret=os.environ.get("INTEGRY_APP_SECRET"),
)

slack_post_message = await integry.functions.get("slack-post-message", user_id)

messages = [{"role": "user", "content": "Say hello to my team on slack."}]

response = await litellm.acompletion(
    model="gpt-3.5-turbo-1106",
    messages=messages,
    tools=[slack_post_message.get_litellm_tool()],
    tool_choice="auto",
)

await handle_litellm_tool_calls(response, user_id, [slack_post_message])

8. Mistral AI

from mistralai import Mistral
from integry import Integry
from integry import handle_mistralai_tool_calls

user_id = "your user's ID"

api_key = os.environ.get("MISTRAL_API_KEY")
model = "mistral-large-latest"
client = Mistral(api_key=api_key)


# Initialize the client
integry = Integry(
    app_key=os.environ.get("INTEGRY_APP_KEY"),
    app_secret=os.environ.get("INTEGRY_APP_SECRET"),
)

slack_post_message = await integry.functions.get("slack-post-message", user_id)

messages = [{"role": "user", "content": "Say hello to my team on slack."}]

response = await client.chat.complete_async(
    model=model,
    messages=messages,
    tools=[slack_post_message.get_mistralai_tool()],
    tool_choice="auto"
)

await handle_mistralai_tool_calls(response, user_id, [slack_post_message])

Prediction

import os
from integry import Integry

user_id = "your user's ID"

# Initialize the client
integry = Integry(
    app_key=os.environ.get("INTEGRY_APP_KEY"),
    app_secret=os.environ.get("INTEGRY_APP_SECRET"),
)

# Get the most relevant function
predictions = await integry.functions.predict(
    prompt="say hello to my team on Slack", user_id=user_id, predict_arguments=True
)

if predictions:
    function = predictions[0]
    # Call the function
    await function(user_id, function.arguments)

Pagination

List methods are paginated and allow you to iterate over data without handling pagination manually.

from integry import Integry

user_id = "your user's ID"

integry = Integry(
    app_key=os.environ.get("INTEGRY_APP_KEY"),
    app_secret=os.environ.get("INTEGRY_APP_SECRET"),
)

async for function in integry.functions.list(user_id):
    # do something with function
    print(function.name)

If you want to control pagination, you can fetch individual pages by using the cursor returned by the previous page.

from integry import Integry

user_id = "your user's ID"

integry = Integry(
    app_key=os.environ.get("INTEGRY_APP_KEY"),
    app_secret=os.environ.get("INTEGRY_APP_SECRET"),
)

first_page = await integry.apps.list(user_id)

second_page = await integry.apps.list(user_id, cursor=first_page.cursor)