
Wonder Pill makes assumptions visible before you pick an idea
A useful brainstorming aid when the problem is still negotiable. Inspect the assumptions and rejected branches; don't mistake an interesting map for a tested plan.
↗ara-mkr / Wonder PillTL;DR: Wonder Pill, by ara-mkr, is a brainstorming skill for Claude. Its useful move is to make assumptions visible before offering alternative directions. Read it when you are still deciding what to build, not when you need an implementation plan by lunchtime.
Read the branch you wanted to delete
The repository describes a map, an orientation paragraph and written provocations. The skill source ties each alternative to a named assumption. The point is to keep the branches visible, not rush to a winner.
My bet on where the value sits: the branch you are ready to throw away. Read it before deleting it. You may find a requirement everyone accepted without discussing it — or a real constraint the brainstorming has conveniently wished away. Either is more useful than a longer list of ideas.
Take an agent that must finish a whole report before review. Why the whole report? Try reviewing one evidence-backed section first, while changing direction is still cheap. That is my illustrative application, not a result from this repository. The test is whether earlier review catches enough bad assumptions to earn the interruption.
Who should save this
Save this if you are an engineer or product lead with a fuzzy brief. Bring one real decision and keep the hard constraints visible. If the map finds a question you had skipped, it has earned your attention. If it only adds branches, you have more to read, not more to act on.
This is an exploration aid, not a feasibility check or evidence of better creativity. I reviewed the README and skill instructions on September 10; I haven't installed it or run a comparative trial. Read downloaded skills before giving them access to your workspace. An interesting map doesn't make their instructions trustworthy.
Once a branch becomes something you intend to ship, switch to the owner-and-expiry rule for agent-built apps. Exploring the idea and taking responsibility for it are different jobs.
Save the assumption you need to test, not just the idea you liked.


