EPIC - Additional OTTB agent implementations

[mmurad2]

Description

This Epic focuses on providing developers with useful single agent implementations that can be useful on their own with some customization OR that can be plugged into a multi-agent workflow to solve a complex use case.

We currently have ReActAgent as a core building block. There are other components that should be added to the developer's toolkit

Initiative / goal

This Epic supports the larger initiative of minimizing the friction with getting to value with the framework.

Hypothesis

Additional OTTB agent implementations to consider:

• Evaluator - An agent that evaluates the output of a multi-agent workflow
• Critique - An agent that provides feedback on the interim steps that agent or multi-agent system follows to help with self-correction.
• ToolCallingAgent - A simple agent implementation that takes a prompt and set of tools (no ReAct). Possible intersection here with supporting native tool calling.

Scope

Final scope TBC based on user feedback

Timeline

End of March TBC