Router skill that loads the right brand knowledge chunks for any content task. Replaces loading individual brand-voice, style-guide, audience-personas, and product-knowledge skills. Uses retrieval rules to optimize context usage.
This skill loads brand knowledge from brand-system/ chunks based on the task at hand. It replaces loading full skills by reading only the chunks you need.
<essential_principles>
These are non-negotiable. They apply to every piece of MyVault content, regardless of task type.
MyVault is a tool, not a personality.
| Do | Never |
|---|---|
| "MyVault found 12 duplicate photos" | "I found 12 duplicate photos" |
| "MyVault organized your vacation photos" | "I organized your vacation photos" |
| "Your vault is ready" | "I'm ready for you" |
Agents speak as tools, not friends. "Insurance Agent analyzed your policies" not "I analyzed your policies."
<intake> </intake> <routing> </routing>Security messaging reassures. It never scares.
| Do | Don't |
|---|---|
| "Your photos stay private" | "Protect yourself from hackers" |
| "We can't see your photos. By design." | "In a world of data breaches..." |
| "Privacy built in from day one" | "Before it's too late" |
Read it out loud. Does it sound like a trusted friend who's great with technology? Does it build confidence or create fear? Is MyVault a helpful tool or a fake personality? Does it explain AI decisions or hide behind "magic"?
</essential_principles>
Determine the task type from context:
If the task type is clear from context, proceed directly to loading. If ambiguous, ask.
After determining the task type, read brand-system/_retrieval-rules.yaml and load chunks accordingly:
brand-system/_retrieval-rules.yamltask_profiles entryalways_loadload_if_relevant based on the specific topic, audience, or contextRead each chunk file from brand-system/ using the directory structure: brand-system/{domain}/{chunk_id}.md (e.g., brand-system/voice/voice-core.md).
| Task | Always Load (minimum) |
|---|---|
| Blog post | voice-core, positioning-core, constraints-messaging, constraints-style |
| Newsletter | voice-core, voice-by-channel, positioning-core, constraints-messaging, constraints-style |
| voice-core, voice-by-channel, constraints-messaging, constraints-style | |
| X post | voice-core, constraints-messaging, constraints-style |
| Website copy | voice-core, positioning-core, brand-about, terminology, constraints-messaging, constraints-style |
| Product copy | voice-core, product-core, product-agents, terminology, constraints-messaging, constraints-style |
| Help docs | voice-core, terminology, product-agents, constraints-style |
| Agent UI | voice-core, terminology, constraints-messaging |
| Style edit | voice-core, constraints-messaging, constraints-style, terminology |
| Brainstorm | positioning-core, positioning-differentiators |
After loading the chunks, apply their guidance throughout your work.
<reference_index>
Positioning: positioning-core, positioning-differentiators Voice: voice-core, voice-translations, voice-by-channel Audience: persona-families, persona-professionals, persona-hnw, persona-anti Identity: brand-about, visual-system, terminology Product: product-core, product-agents, product-v1-scope Guardrails: constraints-messaging, constraints-style
All chunks live in brand-system/{domain}/{chunk_id}.md. The _manifest.yaml is a documentation index for external tools and portability.
</reference_index>