> What started as a community experiment is becoming infrastructure. The developers behind Ralph and Taskmaster figured out something real. Now the platforms are catching up.
> That’s usually how it goes. The practitioners find the patterns first. Then the patterns become features.
This is the scariest thing atm with the fast pacing of these things. As capabilities increase, everything you've spent time on building (w/ scaffolding, tooling, etc) gets "merged" into the all-you-can-prompt solution that the big labs provide. If your previous work has no differentiation, it's very hard to provide additional value / monetise it. And it's hard to know what will be differentiation or what will get eaten up.
It's that sci-fi story trope of the colony ship that gets overtaken by a new generation engine, and when they reach their planet they find a thriving colony there already. But with software :)
The complexity aspect of the planning is one aspect I’d been missing. It fits nicely next to “list assumptions and then check them” and “finish one f#%^ing thing at a time” advice that I give to human teams which is also a struggle for agents.
What both Ralph Loops and Taskmaster miss: Vetted planning processes to ensure you thought through all the details that need to be in the tasks. Did the plans generated by the tools think through all the security, data modeling, performance gaps, and so many other checklists of considerations?
If it's not baked in, the AI may be skipping over it. You'd have no idea until you run into something ugly.
Yes agree that’s why my PRD gets vetted by another agent before I even start breaking it down. Takes all dimensions you mentioned. Happy to share more about this if other are interested
> Without proper task sequencing, agents kept stepping on each other. Which leads to the obvious question: what exactly does each tool give you?
In Gas Town, they “agent their way out” - it has a different type of worker that merges work and if needed can creatively rewrite completed work to suit the new state of a fast moving codebase.
So- sometimes even folks who build agent tools miss chances to take themselves out of the loop! But it’s one of the most powerful ways to scale yourself.
9 comments
[ 2.5 ms ] story [ 26.5 ms ] thread> That’s usually how it goes. The practitioners find the patterns first. Then the patterns become features.
This is the scariest thing atm with the fast pacing of these things. As capabilities increase, everything you've spent time on building (w/ scaffolding, tooling, etc) gets "merged" into the all-you-can-prompt solution that the big labs provide. If your previous work has no differentiation, it's very hard to provide additional value / monetise it. And it's hard to know what will be differentiation or what will get eaten up.
It's that sci-fi story trope of the colony ship that gets overtaken by a new generation engine, and when they reach their planet they find a thriving colony there already. But with software :)
You can have a "Ralph Loop [that] implements the Ralph Wiggum technique" but you can't have a "Ralph Wiggum loop".
https://github.com/anthropics/claude-plugins-official/commit...
If it's not baked in, the AI may be skipping over it. You'd have no idea until you run into something ugly.
In Gas Town, they “agent their way out” - it has a different type of worker that merges work and if needed can creatively rewrite completed work to suit the new state of a fast moving codebase.
So- sometimes even folks who build agent tools miss chances to take themselves out of the loop! But it’s one of the most powerful ways to scale yourself.