docs(website): correct the auto-lint hook's illustrated output (#338)

The quick-start and hooks reference showed the lint hook printing
`SAFEWORD: Linting...`. That's doubly wrong: the `SAFEWORD:` prefix was
removed from all hook output (#332), and the hook never emitted a
"Linting..." banner — post-tool-lint auto-fixes quietly and surfaces only
remaining errors (to the agent) or tool/pack warnings.

Describe what users actually see: lint runs automatically and fixes what
it can; remaining problems are surfaced to the agent.

Co-authored-by: T <t@t.com>
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>

Author: 3 people

packages/website/src/content/docs/getting-started/quick-start.mdx

@@ -53,7 +53,7 @@ Open your project and ask your agent to build something real:
     Watch what happens:
 
     1. Your agent writes the endpoint
-    2. Lint runs automatically — you'll see `SAFEWORD: Linting...` in the output
+    2. Lint runs automatically on the new code, auto-fixing what it can
     3. If there are issues, your agent fixes them before you even see the code
     4. When you stop the session, a quality review runs automatically
 

packages/website/src/content/docs/reference/hooks-and-skills.mdx

@@ -117,7 +117,7 @@ After your agent writes or modifies a file, the right linter runs:
 
 If a linter isn't installed, it's silently skipped.
 
-**What you see:** `SAFEWORD: Linting...` appears in output. If there are issues, your agent sees them and fixes immediately.
+**What you see:** Usually nothing — the linter auto-fixes quietly. If problems remain, your agent sees them and fixes them immediately.
 
 ---