I like having a sub agent for "pre-commit fixer" that the main agent uses before committing so it can pass pre-commit checks. I find it useful because it uses less of the main agents context.
In contrast, I find the scope of the ones in this repository a little weird. The claude main agent does great with CLAUDE.md context, so does it really make sense for it to delegate different tasks to a frontend and backend agent and coordinate between the two? Especially when the additional context in the frontend agent definition is a listing of what libraries and patterns are used in the repo?
atonse · 6h ago
I try to follow all this AI stuff and use AI more (including Claude Code and Cursor) than most people in my professional network ... and I still somehow missed the whole "Subagent" thing... and now there are already catalogs full of them?
It truly feels futile to stay on top of all the new stuff coming out.
bananapub · 6h ago
> This collection features battle-tested subagent definitions following best practices and industry standards
Amazing there’s been time for battles and best practices and industry standards in the like weeks since Anthropic added this feature.
I really had not anticipated that LLMs would claim rational use of adjectives and superlatives before they took my job.
In contrast, I find the scope of the ones in this repository a little weird. The claude main agent does great with CLAUDE.md context, so does it really make sense for it to delegate different tasks to a frontend and backend agent and coordinate between the two? Especially when the additional context in the frontend agent definition is a listing of what libraries and patterns are used in the repo?
It truly feels futile to stay on top of all the new stuff coming out.
Amazing there’s been time for battles and best practices and industry standards in the like weeks since Anthropic added this feature.
I really had not anticipated that LLMs would claim rational use of adjectives and superlatives before they took my job.