In order to provide a single location for Red Hat ET's point of view on best practices for building cloud native agents, I propose changing the structure of this repo to accommodate a wider scope.
Suggestion:
-
Maintain the current Github PR Review Agent demo as our conical example. Move all code to deploy it under /demos/github_review_agent.
-
Add an additional directory /best_practices that includes additional subdirectories focused on each individual component of a cloud native agent application. These subdirectories will include working demos for the open source projects we evaluate for each component. Initial suggestions for list of subdirectories:
- Agent Frameworks
- Memory
- Tools
- Multi-Agent Systems
- Deployment
- Security
- Registries and Discovery
- Observability
- Evaluation
- ...
My hope is that this new structure will allow for greater team collaboration and velocity as we work through identifying the challenges and best practices associated with deploying production agents. I would also like this repo to serve as an easy to consume utility for people outside our group looking to learn about how best to build and deploy agents.
In order to provide a single location for Red Hat ET's point of view on best practices for building cloud native agents, I propose changing the structure of this repo to accommodate a wider scope.
Suggestion:
Maintain the current Github PR Review Agent demo as our conical example. Move all code to deploy it under
/demos/github_review_agent.Add an additional directory
/best_practicesthat includes additional subdirectories focused on each individual component of a cloud native agent application. These subdirectories will include working demos for the open source projects we evaluate for each component. Initial suggestions for list of subdirectories:My hope is that this new structure will allow for greater team collaboration and velocity as we work through identifying the challenges and best practices associated with deploying production agents. I would also like this repo to serve as an easy to consume utility for people outside our group looking to learn about how best to build and deploy agents.