dynamiq — AI Agent 工作流中文教程

Create a workflow to run the agents sequentially based on dependencies.

Getting Started

Ready to dive in? Here's how you can get started with Dynamiq:

Installation

First, let's get Dynamiq installed. You'll need Python, so make sure that's set up on your machine. Then run:

pip install dynamiq

Or build the Python package from the source code:

git clone https://github.com/dynamiq-ai/dynamiq.git
cd dynamiq
poetry install

Setup your LLM (Large Language Model) Node

llm = OpenAI( id="openai", # Unique identifier for the node connection=OpenAIConnection(api_key="OPENAI_API_KEY"), # Connection using API key model="gpt-4o", # Model to be used temperature=0.3, # Sampling temperature for the model max_tokens=1000, # Maximum number of tokens in the output prompt=prompt # Prompt to be used for the model )

Setup your LLM

llm = OpenAI( id="openai", connection=OpenAIConnection(api_key="OPENAI_API_KEY"), model="gpt-4o", temperature=0.3, max_tokens=1000, )

Setup your LLM

llm = OpenAI( connection=OpenAIConnection(api_key="OPENAI_API_KEY"), model="gpt-4o", temperature=0.1, )

Setup your LLM

llm = OpenAI( connection=OpenAIConnection(api_key="OPENAI_API_KEY"), model="gpt-4o", temperature=0.1, )

first_agent = Agent( name="Expert Agent", llm=llm, role="Professional writer with the goal of producing well-written and informative responses.", # Role of the agent id="agent_1", max_loops=5 )

second_agent = Agent( name="Poetic Rewriter Agent", llm=llm, role="Professional writer with the goal of rewriting user input as a poem without changing its meaning.", # Role of the agent id="agent_2", depends=[NodeDependency(first_agent)], # Set dependency on the first agent input_transformer=InputTransformer( selector={"input": f"${[first_agent.id]}.output.content"} # Extract the output of the first agent as input ), max_loops=5 )

Examples

Configuring Two Parallel Agents with WorkFlow

```python from dynamiq import Workflow from dynamiq.nodes.llms import OpenAI from dynamiq.connections import OpenAI as OpenAIConnection from dynamiq.nodes.agents import Agent

The `Workflow` class simplifies setting up and executing a series of nodes in a pipeline.

Configuring Two Sequential Agents with WorkFlow

```python from dynamiq import Workflow from dynamiq.nodes.llms import OpenAI from dynamiq.connections import OpenAI as OpenAIConnection from dynamiq.nodes.agents import Agent

from dynamiq.nodes.node import InputTransformer, NodeDependency

This configuration creates the sequence of states from START -> "generate_sketch" -> "gather_feedback".

orchestrator.add_edge(START, "generate_sketch") orchestrator.add_edge("generate_sketch", "gather_feedback")

Create a workflow to run both agents with the same input

wf = Workflow(flow=Flow(nodes=[agent_first, agent_second]))

Run the workflow with an input

result = wf.run( input_data={"input": "How are sin(x) and cos(x) connected in electrodynamics?"}, )

Without a workflow, you would need to run `first_agent`, collect its output,

and then manually pass that output as input to `second_agent`. The workflow automates this process.

wf = Workflow() wf.flow.add_nodes(first_agent) wf.flow.add_nodes(second_agent)

wf = Workflow(flow=Flow(nodes=[agent_first, agent_second]))

Run the workflow with an input

result = wf.run( input_data={"input": "How are sin(x) and cos(x) connected in electrodynamics?"}, )

Initialize the RAG retrieval workflow

retrieval_wf = Workflow()

Equivalent alternative way to define the workflow:

Equivalent alternative way to define the workflow:

Define the first agent: a question answering agent

first_agent = Agent( name="Expert Agent", llm=llm, role="Professional writer with the goal of producing well-written and informative responses.", id="agent_1", max_loops=5 )