Brand Voice Frameworks

Generate marketing copy, landing pages, email sequences, and social media posts with consistent brand voice. Use when asked to "write copy", "draft a landing...

Security Scan

This is an instruction-only copywriting helper whose requirements and runtime instructions match its stated purpose and contain no obvious incoherent or risky behavior.

Purpose & Capability

Name/description match the instructions: the skill is a framework-driven copy generator that uses local brand reference files. It does not request unrelated credentials, binaries, or services.

Instruction Scope

SKILL.md instructs the agent to read brand reference files from the local references/ directory and apply copy frameworks. It does not direct the agent to read unrelated system files, call external endpoints, or exfiltrate data.

Install Mechanism

No install spec and no code files — instruction-only. Nothing will be written to disk or downloaded by the skill itself.

Credentials

No environment variables, credentials, or config paths are required. The only inputs are the included references/ files and user prompts, which are appropriate for the task.

Persistence & Privilege

always is false and model invocation is allowed (platform default). The skill does not request persistent system-level privileges or modify other skills.

Assessment

This skill is low-risk because it's instruction-only and uses local brand files, but take these precautions before installing: 1) Review any brand files you place in references/ — the agent will read them, so do not store secrets or credentials there. 2) Verify factual claims in generated copy (the skill encourages specific numbers and social-proof statements; do not publish fabricated metrics). 3) Note provenance: the skill has no homepage and unknown source — if you require vendor accountability, prefer skills from known publishers. 4) Because the agent may be invoked normally by the model, review outputs for compliance, legal/marketing accuracy, and confidentiality before use.

Brand Voice

Marketing copy generation with framework-driven consistency. User-first, transformation-focused, always with a clear CTA.

Instructions

Step 1: Identify the Brand

Determine which brand this is for, then load its reference file from references/.

Each brand file should contain: identity, tagline, ICP (ideal customer profile), customer psychology, tone, key messages, and anti-patterns.

See references/cyberdyne-systems.md for an example brand file.

Step 2: Pick the Right Framework

Match the copy framework to what you're writing:

Content Type	Framework	When to Use
Landing page hero	PAS (Problem → Agitate → Solve)	When the audience feels the pain already
Email sequence	AIDA (Attention → Interest → Desire → Action)	Nurturing cold leads to conversion
Social posts	Hook → Value → CTA	Short-form engagement
Feature announcement	FABE (Feature → Advantage → Benefit → Evidence)	When you have proof/numbers
Comparison/competitive	BAB (Before → After → Bridge)	Showing transformation
Testimonial page	Social proof cascade	When you have customer stories

Step 3: Apply Copy Principles

1. User-first voice: "You can..." not "Acme allows..."
2. Specificity sells: "200 attendees" beats "large events"
3. Transformation > features: What does their life look like after?
4. Social proof: Real humans, real results
5. Clear next step: Every piece has one CTA
6. Hook first: Reason to care in the first second

Step 4: Customer Psychology

Always ask before writing:

• "Who is this actually for?" — no "generic user"
• "What's the transformation?" — features → life after
• "What's the hook?" — find it before writing

Step 5: Deliver with Variants

Every deliverable includes:

1. Headline (2-3 A/B variants)
2. Subheadline (supports the hook)
3. Body copy (transformation-focused)
4. CTA (single, clear action)
5. Hypothesis — why this version should work

Step 6: Platform Adaptation

Adapt the same message for different platforms:

Platform	Constraints	Style
Landing page	No length limit	Full narrative, multiple sections
Email	Subject line < 50 chars	Personal, conversational, one CTA
X/Twitter	280 chars	Punchy, hook-first, link in bio
LinkedIn	3000 chars	Professional, story-led, insight-driven
Discord	No markdown tables	Bold + bullets, casual tone

"This / Not That" — Universal Rules

✅ This	❌ Not That
"You'll have a site live in 10 minutes"	"Our platform enables rapid deployment"
"200 parents signed up in one weekend"	"Our solution drives high adoption rates"
Specific numbers and names	Vague superlatives ("best", "powerful", "revolutionary")
Customer language	Product language
One CTA per piece	Multiple competing CTAs
Earned confidence	Hype or desperation

Adding New Brands

Create a new file in references/ following this structure:

# [Brand Name] Brand Guide

## Identity
- **Domain:** example.com
- **Tagline:** "Your tagline here"
- **ICP:** Who is the ideal customer?
- **Market:** Geographic or demographic focus

## Customer Psychology
- **Primary driver:** What motivates the purchase?
- **Fear:** What are they afraid of?
- **Aspiration:** What do they want to become?

## Voice & Tone
- **Personality:** [e.g., confident but approachable]
- **Register:** [e.g., casual professional]
- **Unique phrases:** [brand-specific language]

## Key Messages
1. [Primary value prop]
2. [Secondary value prop]
3. [Social proof angle]

## Anti-Patterns (Never Do This)
- [Brand-specific mistakes to avoid]

Gotchas

• Generic copy is worse than no copy — If you can swap in any competitor's name and it still works, the copy is too generic. Rewrite with product-specific details.
• Framework mismatch kills conversion — PAS on a feature announcement feels manipulative. FABE on a cold email feels robotic. Match the framework to the content type.
• Don't mix brand voices — Each brand has its own personality. Writing Brand A copy in Brand B's voice confuses both audiences.
• CTA competition — Two CTAs on one page split attention and reduce conversion. Pick the one that matters most.
• Load the brand reference every time — Don't rely on memory. Brand details drift. Read the file.