Skip to main content

Quick Start Guide

This guide will help you get started with Astonish by creating and running your first agentic workflow.

Prerequisites​

Before you begin, make sure you have:

  1. Installed Astonish
  2. Configured an AI provider

Creating Your First Agent​

Astonish comes with a built-in agent creator that helps you design new agentic workflows. Let's use it to create a simple agent.

Step 1: Run the Agent Creator​

astonish agents run agents_creator

This will start the agent creator, which will guide you through the process of creating a new agent.

Step 2: Describe Your Agent​

When prompted, provide a description of what you want your agent to do. For example:

I want to create an agent that reads a file, extracts key information, and summarizes it for the user.

The agent creator will use this description to generate a YAML configuration for your agent.

Step 3: Review and Refine​

The agent creator will present you with a YAML configuration for your agent. Review it and provide feedback if needed. You can:

  • Approve the configuration if it looks good
  • Provide feedback to refine the configuration

Step 4: Save the Agent​

Once you're satisfied with the configuration, the agent creator will save it to your config directory. It will provide you with the name of the saved agent.

Running Your Agent​

Now that you've created an agent, let's run it:

astonish agents run your_agent_name

Replace your_agent_name with the name of the agent you just created.

Example: File Summarizer Agent​

Let's walk through an example of creating and running a file summarizer agent:

Step 1: Run the Agent Creator​

astonish agents run agents_creator

Step 2: Describe Your Agent​

I want to create an agent that reads a file, extracts key information, and summarizes it for the user.

Step 3: Review and Refine​

The agent creator will generate a YAML configuration similar to this:

description: An agent that reads a file, extracts key information, and summarizes it for the user
nodes:
  - name: get_file_path
    type: input
    prompt: |
      Please enter the path to the file you want to analyze:
    output_model:
      file_path: str

  - name: read_file_content
    type: llm
    system: |
      You are a file reading assistant.
    prompt: |
      Read the contents of the file at path: {file_path}
    output_model:
      file_content: str
    tools: true
    tools_selection:
      - read_file

  - name: extract_key_info
    type: llm
    system: |
      You are an AI assistant specialized in extracting and summarizing key information from text.
    prompt: |
      Analyze the following file content and extract the core information:

      {file_content}

      Provide a concise summary of the key points.
    output_model:
      summary: str

  - name: present_summary
    type: llm
    system: |
      You are a helpful AI assistant presenting information to users.
    prompt: |
      Present the following summary to the user in a clear and engaging manner:

      {summary}
    output_model:
      final_response: str
    user_message:
      - final_response

flow:
  - from: START
    to: get_file_path
  - from: get_file_path
    to: read_file_content
  - from: read_file_content
    to: extract_key_info
  - from: extract_key_info
    to: present_summary
  - from: present_summary
    to: END

Step 4: Save the Agent​

The agent creator will save the configuration and provide you with the name of the saved agent, for example, file_summarizer.

Step 5: Run the Agent​

astonish agents run file_summarizer

The agent will:

  1. Ask you for the path to a file
  2. Read the file using the read_file tool
  3. Extract key information from the file content
  4. Present a summary to you

Visualizing the Agent Flow​

To better understand how your agent works, you can visualize its flow:

astonish agents flow file_summarizer

This will print an ASCII representation of the agent's flow, showing how the nodes are connected.

Next Steps​

Now that you've created and run your first agent, you can:

  1. Learn more about Agentic Flows to understand how Astonish works
  2. Explore the YAML Configuration to customize your agents
  3. Check out the Tutorials for more advanced examples
  4. Learn about Tools to extend your agents' capabilities