@cavuno/board 1.44.0 → 2.0.1

package/skills/cavuno-board-smoke-test/SKILL.md

@@ -29,19 +29,10 @@ named in its summary and still needs the manual checks below. Use the
 rest of this skill for the behavioral checks doctor does not automate
 (auth flows, gating semantics, production build).
 
-Add `--sandbox` to also run the write probes (register → login → apply →
-alert-signup). They are refused unless the board the `pk_` resolves is
-THE platform sandbox (`sandbox: true` in the board context — set only by
-the platform's sandbox seed and stripped from any client settings write)
-— never point them at a real tenant. The
-email-delivery check additionally needs `RESEND_API_KEY` (a platform
-operator credential) and SKIPs loudly without it.
-
-Rate limits are real on the sandbox: the auth endpoints share a
-10-requests/min bucket and register allows 5 signups per 15 minutes per
-IP, and one probe run consumes 4 requests + 1 signup. Rapid reruns will
-start failing with HTTP 429 — that's the limiter working, not a broken
-board; wait a few minutes.
+Doctor is non-destructive. It validates configuration, the public Board API,
+installed skills, and—when `--frontend` is provided—rendered pages, sitemaps,
+robots rules, and JobPosting structured data. It never creates users,
+applications, or alerts and needs no Cavuno operator credentials.
 
 ## 1 — Probe the API directly (before blaming app code)
 

package/skills/flavors/tanstack-start/SKILL.md

@@ -73,8 +73,7 @@ let inflight: Promise<void> | null = null;
 export async function withFreshSession<T>(
   getAccessToken: () => string,
   getRefreshToken: () => string,
-  run: (accessToken: string) => Promise<T>,
-): Promise<T> {
+  run: (accessToken: string) => Promise<T>): Promise<T> {
   try {
     return await run(getAccessToken());
   } catch (err) {

package/skills/manifest.json

@@ -1,5 +1,5 @@
 {
-  "version": "1.44.0",
+  "version": "2.0.1",
   "skills": [
     {
       "name": "cavuno-board-account",