refactor(checkpoint): deduplicate step rules against global step rules

Steps 2–4 now reference Global Step Rules in SKILL.md instead of
restating path:line format, front-load, and silence rules locally.
Step-specific rules (concern-based org, design judgment, risk
awareness, experiential testing) are preserved.

src/bmm-skills/4-implementation/bmad-checkpoint-preview/step-02-walkthrough.md

@@ -2,11 +2,9 @@
 
 Display: `Orientation → [Walkthrough] → Detail Pass → Testing`
 
-## RULES
+## Follow Global Step Rules in SKILL.md
 
 - Organize by **concern**, not by file. A concern is a cohesive design intent — e.g., "input validation," "state management," "API contract." One file may appear under multiple concerns; one concern may span multiple files.
-- Every code reference uses clickable `path:line` format per the standing rule.
-- **Front-load then shut up.** Present the entire walkthrough in a single message. Do not ask questions mid-walkthrough. Do not pause between concerns. After presenting, go quiet — the human reads, clicks, navigates at their own pace.
 - The walkthrough activates **design judgment**, not correctness checking. Frame each concern as "here's what this change does and why" — the human evaluates whether it's the right approach for the system.
 
 ## BUILD THE WALKTHROUGH
@@ -81,8 +79,6 @@ When you are done with the walkthrough:
 - **"I've seen enough"** — tell me what to do about this {change_type} and we wrap up
 ```
 
-**Do NOT speak again until the human responds.**
-
 ## NEXT
 
 When the human signals readiness for the next phase, read fully and follow `./step-03-detail-pass.md`

src/bmm-skills/4-implementation/bmad-checkpoint-preview/step-03-detail-pass.md

@@ -2,10 +2,8 @@
 
 Display: `Orientation → Walkthrough → [Detail Pass] → Testing`
 
-## RULES
+## Follow Global Step Rules in SKILL.md
 
-- Every code reference uses clickable `path:line` format per the standing rule.
-- **Front-load then shut up.** Present all findings in a single message. Do not drip-feed.
 - The detail pass surfaces what the human should **think about**, not what the code got wrong. Machine hardening already handled correctness. This activates risk awareness.
 - The LLM detects risk category by pattern. The human judges significance. Do not assign severity scores or numeric rankings — ordering by blast radius (below) is sequencing for readability, not a severity judgment.
 - If no high-risk spots exist, say so explicitly. Do not invent findings.
@@ -82,8 +80,6 @@ You've seen the design and the risk landscape. From here:
 - **"I've seen enough"** — tell me what to do about this {change_type} and we wrap up
 ```
 
-**Do NOT speak again until the human responds.**
-
 ## TARGETED RE-REVIEW
 
 When the human says "dig into [area]" (e.g., "dig into the auth changes", "dig into the schema migration"):

src/bmm-skills/4-implementation/bmad-checkpoint-preview/step-04-testing.md

@@ -2,10 +2,8 @@
 
 Display: `Orientation → Walkthrough → Detail Pass → [Testing]`
 
-## RULES
+## Follow Global Step Rules in SKILL.md
 
-- Every code reference uses clickable `path:line` format per the standing rule.
-- **Front-load then shut up.** Present all suggestions in a single message. Do not drip-feed.
 - This is **experiential**, not analytical. The detail pass asked "did you think about X?" — this says "you could see X with your own eyes."
 - Do not prescribe. The human decides whether observing the behavior is worth their time. Frame suggestions as options, not obligations.
 - Do not duplicate CI, test suites, or automated checks. Assume those exist and work. This is about manual observation — the kind of confidence-building no automated test provides.
@@ -73,8 +71,6 @@ You've seen the change and how to verify it. From here:
 - or just ask anything
 ```
 
-**Do NOT speak again until the human responds.**
-
 ## WRAP-UP
 
 When the human says "I've seen enough" or signals they're done: