The skill-as-product pattern is quietly becoming the default.
§ i Top-line signal
The market is still deciding what a Claude app looks like. Portable skill bundles — small, copyable, zero infrastructure — are winning by default because they are the lightest thing that works. Heavier harnesses keep losing to the thing the user already has open. Builders who ride the platform are compounding; builders who compete with it are stuck explaining why their wrapper is worth leaving Claude for.
§ ii Project radar
Jibe’s catalog holds five SKUs across three states — shipping, building, spec’ing. The unifying thesis is tools with a point of view. The catalog’s job is to make the point of view visible before any single tool is. On that job, it is on track. The next decision worth closing is not which tool next — it is which reader each tool is written for, stated plainly on the SKU page.
§ iii Angles worth sharing
- The move from heavy harness to a 98 KB skill file that you actually use is a teachable arc. Most solo projects over-build before they under-build — and the story is most useful told by someone who did both.
- The services-thesis frame — free core, paid customization, in-context sales — is underpriced as a solo-builder playbook. Few people are writing about it, and it is a clean counter to the “race to free” narrative everyone else is stuck in.
§ iv Three things to try this week
- Have one conversation with a prospective customization buyer — not to sell, to learn what they would pay to have tuned. Price intuition comes from contact, not spreadsheets.
- Write one long-form piece this week instead of three threads. The compounding is on the essay, not the posts.
- Close one open tab. You know the one.