ck-framelisted
Install: claude install-skill dbbz/C-K-studio
# Framing the root concept
The single highest-leverage step. A root that fails the undecidability test produces
a tree of obvious answers no matter how well the rest of the session is run.
Read `${CLAUDE_PLUGIN_ROOT}/references/GRAMMAR.md` first.
## Procedure
### 1. Seed K before writing C
You cannot test undecidability against an empty K. Four to eight propositions defining
what is currently known and currently constraining is plenty.
**Research and present them; don't ask for them.** Supplying propositions from a blank
page is hard work and it is your job, not theirs. Correcting a list you put in front of
them is easy, fast, and produces sharper K — people who cannot generate a constraint
will immediately tell you when one is wrong.
```bash
ck add-k <session> --text "..." --status true --source "..." --op seed \
--summary "Seeded the constraint everyone treats as a law of nature."
```
Include the constraint that makes the problem interesting. In the worked example
that's hull speed: the thing everyone treats as a law of nature.
### 2. Convert the user's framing into concept form
People arrive with goals, requirements, or solutions. None of those are concepts.
| They say | What it is | Reformulate to |
|---|---|---|
| "Cut delivery cost 30%" | a target | "There exists a delivery system with no marginal cost per drop." |
| "We should use drones" | a solution | Ask what property of delivery drones break. Then make *that* the concept. |
| "Our onboarding is too slo