docs(skill): align debug/design prompts with shipped browser commands + changeset

Only reference implemented browser verbs (open/eval/snapshot/screenshot/
grab); flag console/network/audit as upcoming. Adds the changeset.

Author: aidenybai

.changeset/react-skill-umbrella.md

@@ -0,0 +1,5 @@
+---
+"react-doctor": minor
+---
+
+Expand the agent skill into a routing umbrella (`/react-doctor`): ten always-on React baseline rules plus prose routing across doctor (static scan), perf (the DevTools profiling harness), debug, design, and grab. Adds an in-house browser core (`@react-doctor/browser`) that attaches to a running Chrome over CDP (launching one only as a fallback) and a `react-doctor browser` command (open/eval/snapshot/screenshot/grab) that backs the debug, design, and grab jobs. The browser core loads lazily, so playwright never bundles into the CLI's hot path.

skills/react-doctor/references/debug.md

@@ -34,11 +34,14 @@ npx react-doctor browser eval "document.querySelector('[data-testid=cart]')?.tex
 2. **Observe.** Pull the runtime signals around the failure:
 
 ```bash
-npx react-doctor browser console --errors   # console errors + warnings
-npx react-doctor browser network --failed    # 4xx/5xx and failed requests
-npx react-doctor browser snapshot            # accessibility tree + React component tree
+npx react-doctor browser snapshot                          # what actually rendered
+npx react-doctor browser eval "window.__lastError?.stack"  # read state/DOM at the failure
 ```
 
+Dedicated `console` and `network` capture commands (errors, 4xx/5xx, failed
+requests) are landing next; until then, drive the repro and read the relevant
+state or error with `eval`.
+
 3. **Hypothesize** (2 to 3): a thrown error in a specific component, a failed or
    duplicated request, a null/undefined access, a state update on an unmounted
    component, a missing loading/error branch.

skills/react-doctor/references/design.md

@@ -13,12 +13,18 @@ size. Lead with those, then apply craft.
 ```bash
 npx react-doctor browser open http://localhost:3000
 npx react-doctor browser screenshot --out review.png   # what the user actually sees
-npx react-doctor browser audit                          # contrast, target size, a11y findings
 ```
 
-`audit` reports measured issues with the element and the failing value (e.g. a
-3.1:1 contrast ratio on body text, a 28px tap target). Fix those first; they are
-objective and a smarter model cannot dismiss them as opinion.
+Look at the screenshot, then measure specifics with `eval` (computed styles,
+bounding boxes, color values) to get objective numbers rather than opinions:
+
+```bash
+npx react-doctor browser eval "getComputedStyle(document.querySelector('button')).fontSize"
+```
+
+A one-shot `audit` command (axe-based contrast, tap-target, and a11y findings in
+one report) is landing next; until then, measure the checks below with `eval`.
+Lead with the measured issues; a smarter model cannot dismiss them as opinion.
 
 ## What to check