This article provides a comprehensive guide on how to manage agents within projects in AgentRunner. Agents are the core of your automation workflows, and understanding how to manage them effectively is crucial for building and maintaining robust and efficient automation. This article will cover everything from accessing your agents to running, editing, and maintaining them. It will also explain the various actions you can take on your agents, such as version control, log checking, duplication, deletion, and sharing.
Accessing and Navigating Agents
Projects in AgentRunner serve as containers for your agents, providing a structured way to organize and manage your automation workflows. Think of projects as folders that hold one or more agents. Each agent represents a specific sequence of actions or tasks that your automation will perform. Understanding how to navigate and access these flows is the first step in effectively managing them.
Actions Available for Agents
Once you have located an agents within a project, you have access to a range of actions that allow you to manage and interact with the agent. These actions include the ability to edit the agent, run the agent, modify its title and description, manage versions, check logs, duplicate the agent, delete the agent, and share the agent with other users. Each of these actions plays a vital role in the lifecycle of an agent, from its initial creation and testing to its ongoing maintenance and collaboration.
Agent Operations: Editing, Running, and More
AgentRunner offers a range of operations you can perform on agents within a project. These operations are essential for developing, testing, and maintaining your automation workflows. Understanding how to use these operations effectively is critical for maximizing the power of AgentRunner.
Editing and Running Agents
Editing Agents: Only managers and owners can edit an agent. Editing allows you to modify the agent's logic, add or remove nodes, and adjust the agent's parameters. This is where you define the specific actions and decisions that your agent will take.
Running Agents: Anyone with access to the agent can run it. Running a agent executes the defined sequence of actions, allowing you to test and validate your automation logic. Agents are executed immediately upon initiation.
Visual Recommendation: Consider using annotated screenshots to illustrate the edit and run buttons within the AgentRunner interface. This will help users quickly locate these actions.
Additional Agent Management Options
Beyond basic editing and running, AgentRunner provides a suite of additional agent management options to help you maintain and collaborate on your automation workflows.
Edit Title and Description: You can modify the title and description of a agent to better reflect its purpose and functionality. This helps with organization and makes it easier for others to understand the agent's role within the project.
Handle Versions: AgentRunner provides version control, allowing you to clone versions, and change the labels on your agents to determine which one is the stable version and which one is under development, as well as allowing you to create custom labels. This is crucial for managing changes and ensuring that you can always revert to a previous working state.
Check Logs: You can access run logs for each agent, which contain the input, outputs, and results of each individual node inside. This provides valuable insights into the agent's execution and helps you identify and troubleshoot any issues.
Duplicate Agent: You can duplicate the newest version of the agent within the same project. This is useful for creating variations of an existing agent or for experimenting with new ideas without modifying the original agent.
Delete Agent: Deleting an agent is a permanent action and cannot be undone. It's essential to exercise caution when deleting agents and to ensure that you have a backup if needed.
Share Agents: Sharing an agent generates a shortened URL that you can send to other users who have permissions to access the project the agent is in. This allows you to collaborate with others on the agent's development and maintenance.
Advanced Agent Management: Versioning, Logs, and Collaboration
AgentRunner provides advanced features for managing agents, including version control, detailed logging, and collaboration tools. These features are essential for teams working on complex automation projects.
Version Control and Collaboration
Version Control: AgentRunner offers robust version control capabilities. You can create new versions of an agent, remove older versions, and label versions as "stable", "under development (dev)", "legacy" or with a custom label you create. This allows you to track changes, revert to previous versions if needed, and manage the development lifecycle of your agents.
Collaboration: Sharing an agent generates a shortened URL that can be shared with other users who have permission to access the project. Version control ensures that multiple users can work on the same agent without overwriting each other's changes.
Logging and Monitoring
Run Logs: AgentRunner provides detailed run logs for each agent execution. These logs contain information about the input, outputs, and results of each node within the agent. You can view the execution time for each run and for each node individually.
Troubleshooting: The run logs are invaluable for troubleshooting issues within your agents. By examining the inputs, outputs, and results of each node, you can quickly identify the source of errors and diagnose problems.
Important Considerations
When working with agents in AgentRunner, it's essential to be aware of certain considerations and limitations. There are no limits to the number of agents that can be created within a project, but agents cannot be organized into folders or subcategories within a project. When running an agent, ensure that you have the proper API keys for the integrations you want to use.
