Hello,
Going through the flows and documentation, I felt something was missing around integrating the framework into existing projects.
I really value the framework being opinionated about what "good" looks like, but in older projects it's very common that not everything follows those standards, even when the rules are written down.
I know there are ways to make project-specific rules available to Claude agents, but the options feel limiting: either I set them up in memory files (potentially bloating the context of unrelated framework agents), or I resort to manual invocation, when what I'd really love is for the automatically invoked reviewers to actually know those rules.
Bottom line, this is more of an open question, but since you've done a fair amount of research into agent optimization, I think having some guidance on "best practices" for integrating external rules and constraints would be incredibly valuable!
Hello,
Going through the flows and documentation, I felt something was missing around integrating the framework into existing projects.
I really value the framework being opinionated about what "good" looks like, but in older projects it's very common that not everything follows those standards, even when the rules are written down.
I know there are ways to make project-specific rules available to Claude agents, but the options feel limiting: either I set them up in memory files (potentially bloating the context of unrelated framework agents), or I resort to manual invocation, when what I'd really love is for the automatically invoked reviewers to actually know those rules.
Bottom line, this is more of an open question, but since you've done a fair amount of research into agent optimization, I think having some guidance on "best practices" for integrating external rules and constraints would be incredibly valuable!