Welcome to the AgentRunner agent editor, a powerful tool designed to empower anyone to build automation flows, regardless of coding experience. This article will guide you through the features and functionalities of the agent editor, enabling you to create efficient and effective automated processes.
Understanding the Agent Editor Interface
The AgentRunner agent editor provides a visual, intuitive interface for constructing automation workflows. It's designed for everyone, even those without coding experience. You can build automating agents simply with the agent editor. Let's explore the core components of the editor:
Core Components
Nodes: These are the building blocks of your agent. Each node performs a specific function.
Input Node: Gathers necessary information for the agent to run. It accepts text strings. You can name and write a default value for the input nodes to explain their meaning.
Output Node: Displays the final results of the agent. It produces a JSON object that can be viewed and copied in JSON, Markdown, and Text formats.
Text Merger: Combines two input texts into one. Because all data is in text format, there are no data type issues.
Selection: Makes a true/false decision based on specified conditions.
Language Models: Accesses various large language AI models, including Anthropic, Gemini, and OpenAI.
Integrated Apps: Integrates with other applications and websites.
Agents Within the Same Project: Allows you to use other agents in the project as a single node within the current agent. This simplifies complex agents.
Canvas: The main area where you arrange and connect nodes to create your agent.
Connections: Lines that link nodes together, defining the flow of data and execution. Building the agent is as easy as connecting the nodes with lines, eliminating the need for code writing.
Easy Agent Creation
The end-to-end process of creating and using an agent is simple: place the nodes, fill them out with the necessary data, and connect them with lines.
Working with Nodes in the Agent Editor
Nodes are the fundamental units within the AgentRunner agent editor, each serving a distinct purpose in the automated workflow. Understanding how to configure and connect these nodes is crucial to building effective agents. Here's a detailed look at the key node types and their functionalities:
Configuring Nodes
Input Node Configuration:
Naming: Give your input node a descriptive name that indicates the type of information it will receive (e.g., "Customer Name," "Product Description").
Default Value: Provide a default value to explain the expected input. For example, for a "Customer Name" input, the default could be "Enter customer's full name."
Selection Node Configuration:
Left Operand: Connect the text you want to compare to the "left operand" input.
Right Operand: Connect the text you want to compare against to the "right operand" input.
Condition Selection: Choose a condition from the dropdown menu. Available conditions include:
"Contains"
"Not contains"
"Starts with"
"Ends with"
"Is"
"Is not"
"Is empty"
"Is not empty"
True/False Branches: If the condition is true, the agent will follow the "true" branch. If the condition is false, it will follow the "false" branch.
Output Node Configuration:
The output node automatically compiles the final result of the agent into a JSON object.
The output can be viewed and copied in JSON, Markdown, and Text formats, making it easy to integrate with other systems.
Testing Nodes: You can test each node individually as well as the entire agent to identify and resolve any issues.
Easy Connections:
Connect two nodes with a simple line that you drag with your mouse.
Integrating External Resources
To connect to integrated apps:
Go to the Apps menu.
Enter your own API keys for the apps you wish to use.
If an integration isn't available by default, users with the technical expertise to connect their own integration can do so.
Saving, Sharing, and Version Control
AgentRunner provides robust features for saving, sharing, and managing versions of your agents, ensuring you can collaborate effectively and maintain a history of your work.
Agent Management
Automatic Saving:
Agents are saved automatically after every edit, so you never have to worry about losing your progress.
Version Control:
You can create new versions of your agents to preserve specific states or experiment with changes.
Creating a new version duplicates the current agent, allowing you to edit the new version while keeping the original intact.
Version Labeling:
Label your versions with statuses such as "stable," "dev" (in development), and "legacy" to help organize and manage your projects.
Sharing Agents:
Agents can be shared with other users within the AgentRunner platform for collaborative development.
API Connectivity:
Agents can be connected as APIs in external applications and websites, enabling seamless integration with other systems.
