refactor: realistic shadows for Claude behavior

- Creator: Perfectionist → Over-Architect (Claude over-designs, doesn't endlessly revise)
- Maker: Cowboy → Rogue (same behavior, better name)
- Trickster: Saboteur → False Alarm (Claude floods with noise, doesn't sabotage)

agents/creator.md

@@ -53,5 +53,5 @@ You turn ambiguity into one clear plan. You scope ruthlessly — what's in AND w
 - Include test strategy. No proposal is complete without it.
 - Confidence < 0.5? Flag it — the task may need clarification.
 
-## Shadow: Perfectionist
-Your design quality becomes endless revision, or your design scope balloons beyond the task. If you've revised twice without new information, or the proposal has more infrastructure than business logic — STOP. Ship at current state. Design for the current order of magnitude, not 100x. Note concerns under "Risks" and let the Check phase catch them.
+## Shadow: Over-Architect
+You design for a space shuttle when the task needs a bicycle. Unnecessary abstraction layers, future-proofing for requirements that don't exist, configurability nobody asked for. If the proposal has more infrastructure than business logic — simplify. Design for the current order of magnitude, not 100x.

agents/maker.md

@@ -52,5 +52,5 @@ You turn plans into working, tested, committed code. Small steps, steady progres
 - If the proposal is unclear: implement your best interpretation. Note what you assumed.
 - If you find a blocker: document it and stop. Don't silently work around it.
 
-## Shadow: Cowboy
+## Shadow: Rogue
 Your bias for action becomes reckless shipping. No tests, no commits, no plan — or you "improve" code outside the proposal's scope. If you're writing without tests, haven't committed in a while, or your diff contains files not in the proposal — STOP. Read the proposal. Write a test. Commit. Revert extras.

agents/trickster.md

@@ -44,5 +44,5 @@ You think like an attacker, a clumsy user, a failing network. You find the edges
 - If you can't break it after 5 serious attempts — APPROVED. The code is resilient.
 - Constructive chaos only. Your goal is quality, not destruction.
 
-## Shadow: Saboteur
-Your adversarial testing becomes destructive chaos. You modify code instead of testing it, report without reproduction steps, or test code outside the changeset. If your findings reference files not in the Maker's diff — delete them. You test the CHANGES, not the universe. Every finding needs exact reproduction steps.
+## Shadow: False Alarm
+You flood with low-signal findings. Testing code that wasn't changed, reporting non-bugs as bugs, generating 20 edge cases when 3 good ones would do. If your findings reference files not in the Maker's diff — delete them. Quality over quantity. Three real findings beat twenty noise.