Announcing the AIAgentWorkflow Editor Beta: Visually Build Your AI Agents

This is evolving daily now.  The second beta is here and it has a dark theme

https://mystry-geek.blogspot.com/2025/08/a-second-feta-for-agent-workflow-editor.html

In the rapidly evolving world of AI, chaining together different models, prompts, and custom code to access legacy data is how powerful applications are born. But as these chains—or workflows—grow, managing them in raw configuration files like JSON can become a complex and error-prone task. A single misplaced comma or a logical error in how you pass parameters can bring your entire system to a halt.

Today, we're excited to take the first step in solving that problem by announcing the beta release of the AIAgentWorkflow Editor, a graphical tool designed to build, manage, and visualize the agents in the AIAgentWorkflow framework.  Look for the config_editor.py file in that github repository.  There is already a config.json with examples to let you get started automating workflows locally today.

What is AIAgentWorkflow?

For those new to the project, AIAgentWorkflow is a powerful Python-based system for executing multi-step automation tasks. The core of the system is the concept of "Agents," which are modular, reusable components that can be chained together. The framework is built around three primary agent types:

• Template Agents: Perfect for dynamic prompt engineering. These agents take your parameters and build the final prompt text to be sent to a language model.  At its core, a Template Agent is a powerful tool for dynamic string generation. While perfect for engineering complex AI agent prompts, its utility extends to any text-based task. Use it to construct dynamic SQL queries, generate precise connection strings for databases, or build the JSON payloads for API calls. From creating formatted reports to assembling shell commands, if your workflow needs to build a precise string from variables, the Template Agent is the clean and reliable tool for the job.

• Proc Agents: For when you need more than just a prompt. Proc Agents transform your workflow from a simple chain of prompts into a true automation engine. They allow you to execute any Python function, giving you unrestricted access to the entire ecosystem of libraries to process data exactly as you need it. This is how you connect your AI to the real world: integrate with legacy databases, interact with any web service, and call any AI API—for text, images, or the next big thing. If you need to fetch, process, or send data, the Proc Agent is your tool. 

• Workflow Agents: This is where the magic happens. Workflow Agents are orchestrators. They are themselves a series of steps that can call other agents—including other workflows! This allows you to build incredibly complex and deeply nested hierarchies of tasks, creating sophisticated applications from simple, reusable building blocks. As you build new workflows, they instantly become reusable agents themselves, enabling you to rapidly compose and scale your automation capabilities.

Introducing the Visual Editor

The AIAgentWorkflow Editor is our solution to managing the complexity of this system. Instead of editing raw configuration files by hand, you can now:

• Create and Manage Agents: Easily add new Template, Proc, or Workflow agents to your configuration.

• Define Agent Properties: Use a simple UI to define the inputs, outputs, and help text for each agent.

• Visually Build Workflows: Add steps to your Workflow Agents by simply clicking on other agents from your list.

• Configure Every Step: A dedicated side panel allows you to configure the parameters for each step in a workflow, including how outputs from one step are mapped as inputs to the next.

• Reduce Errors: By managing the structure for you, the editor helps prevent common syntax and logical errors, letting you focus on the workflow itself.

Our goal is to make the power of the AIAgentWorkflow engine accessible and intuitive, allowing for faster iteration and more ambitious projects.

This is a Beta: Your Feedback is Crucial

We are releasing this editor in an early beta stage because we believe in building with the community. While the core functionality for creating, editing, and managing agents is in place, there are still a few rough edges and minor bugs that require manual workarounds.

For example, the side panel for editing workflow steps can sometimes get into a state where it stops responding after rapidly switching between different steps. The current workaround is simple: Just be sure to click on an open step to close it before you click on the next step you want to edit.

We are actively working to smooth out these interactions and make the entire experience seamless. Your feedback on issues like this is invaluable. By trying out the beta and reporting bugs or suggesting features, you will play a direct role in shaping the future of this tool.

Get Started Today

Ready to dive in? The editor is part of the main project repository.

1. Head over to our GitHub page: https://github.com/BuckRogers1965/AIAgentWorkflow

2. Follow the setup instructions in the README to get the project and its dependencies running.

3. Run the editor and start building your first visual workflow.

4. Submit your feedback, bug reports, and ideas through the "Issues" tab on GitHub.

We are incredibly excited about the potential of the AIAgentWorkflow framework and this new editor. Thank you for your interest, and we can't wait to see what you build with it.