Add shared writing style preferences (#240)

* Add shared writing style preferences

* Address PR review feedback

* Add scoring instructions settings

* Polish writing style inputs

* Clarify do-not-use terms guidance

* Normalize blank writing style overrides

* Refactor writing style resolution to use effective values and enhance blank value handling

docs-site/docs/features/ghostwriter.md

@@ -44,6 +44,8 @@ Global settings affecting generations:
 - `Constraints`
 - `Do-not-use terms`
 
+`Do-not-use terms` are passed as guidance in the prompt. They are not enforced by a hard post-generation filter, so the model should avoid them but may still use them occasionally.
+
 Defaults:
 
 - Tone: `professional`

docs-site/docs/features/settings.md

@@ -15,7 +15,7 @@ It lets you configure:
 
 - LLM provider and models
 - Webhook destinations and secret
-- Display and Ghostwriter defaults
+- Display and writing-style defaults
 - Service credentials and basic auth
 - Reactive Resume project selection
 - Tracer Links readiness verification
@@ -62,15 +62,18 @@ Settings gives you runtime overrides for the key parts of discovery, scoring, ta
 
 - Toggle visa sponsor badge visibility in job lists/details
 
-### Ghostwriter
+### Writing Style
 
 ![Ghostwriter settings section](/img/features/settings-ghostwriter-section.png)
 
+- Pick a preset for a quick starting point
 - Set global writing defaults:
   - Tone
   - Formality
   - Constraints
   - Do-not-use terms
+- These settings apply to Ghostwriter and resume tailoring
+- Do-not-use terms are model guidance, not a guaranteed output filter
 
 ### Reactive Resume
 
@@ -120,6 +123,7 @@ Readiness requires:
 - Set penalty amount
 - Optional auto-skip threshold for low-score jobs
 - Block jobs from companies that match configured keyword tokens
+- Add custom scoring instructions to tell the AI what to weigh more or less
 
 ### Danger Zone