agential-ai/agential

[Feature Request]: Towards Structured, Organized Modules

Closed this issue · 0 comments

Feature Description

Refactoring the entire library.

Currently we have functional -> modules -> strategies -> agent. We also have prompts/benchmark and prompts/agent. For any specific agent and benchmark, files are spread across multiple modules. Is there a way to compartmentalize all relevant logic for an agent into its own specific module/folder?

Plus, we can also simplify the base agent class to just generate().

Oh, another idea, maybe we can find some standard output format @IGiotto12 across the benchmarks for an agent or even across all agents/benchmarks.

Reason

No response