Purpose
Evaluate a piece of draft copy against a brand's documented voice and positioning, flag every off-brand phrase with a specific reason and a compliant rewrite, and return a scored diagnosis so a copy team ships consistent, on-voice work without a director hand-editing every draft.
When to Use
Invoke this skill when the user asks to:
- Check whether a piece of copy is on-brand before it ships
- Review a writer's draft against brand voice guidelines
- Enforce voice consistency across a content or copy team
- Flag off-brand language and get compliant alternatives
- Score or grade a draft against a positioning framework
- Onboard a new writer by showing where their draft drifts from the voice
Inputs Required
Before starting, collect:
- Draft copy: The text to review, with its format noted (email, landing page, ad, blog, social)
- Brand voice guidelines: Documented voice traits, tone descriptors, and any voice charts (e.g., "we are X, not Y")
- Positioning framework: Core value proposition, category, and primary differentiator
- Words/phrases to avoid: Banned terms, competitor language, or overused clichés the brand rejects
- Preferred vocabulary: On-brand terms, product naming conventions, and signature phrases
- Audience: Who the copy addresses and their relationship to the brand
- Channel norms: Any channel-specific rules (character limits, formality shifts, CTA conventions)
If a formal positioning framework is missing, infer positioning from the guidelines and flag the assumption at the top of the review.
Instructions
Step 1: Establish the Voice Baseline
Before reviewing a single line, translate the guidelines into a concrete checklist the review runs against. Extract:
- 3–5 voice traits (e.g., confident, warm, plainspoken) and one behavioral rule for each — what the trait sounds like in a sentence
- The "we are X, not Y" chart — if none exists, construct one from the trait descriptions
- Positioning anchors — the value prop and differentiator every claim should ladder up to
- Banned and preferred vocabulary — a two-column reference to check word choices against
State this baseline in 4–6 bullets so the reviewer's judgments are traceable, not arbitrary.
Step 2: Line-by-Line Flagging
Walk the draft sentence by sentence. For every phrase that drifts, produce a flag with these fields:
- Original: The exact off-brand phrase, quoted
- Issue: Which voice trait, banned term, or positioning anchor it violates
- Severity: High (contradicts positioning or uses a banned term), Medium (weakens the voice or dilutes a claim), Low (minor tone or word-choice drift)
- Rewrite: A compliant alternative in the brand's actual voice — not a generic "improved" version
Flag only genuine drift. Do not flag lines that are already on-brand to appear thorough — note strong on-brand lines separately in Step 4.
Step 3: Voice Dimension Scoring
Score the draft 1–5 on each dimension, with a one-sentence justification tied to specific lines:
- Tone alignment: Does the emotional register match the brand?
- Vocabulary fit: Are word choices on-brand? Any banned terms or missing signature language?
- Positioning consistency: Do claims ladder up to the differentiator, or wander off-message?
- Clarity and rhythm: Does it read the way the brand speaks — sentence length, cadence, directness?
- Audience calibration: Is the formality and assumed knowledge right for this reader and channel?
Compute a total (out of 25) and assign a verdict: Ship as-is (23–25), Minor revisions (18–22), Needs a pass (12–17), Rewrite recommended (below 12).
Step 4: Prioritized Revision List and On-Brand Wins
Close with two lists:
- Fix first: The High-severity flags in priority order, each with the rewrite ready to paste
- On-brand wins: 2–4 lines that nail the voice, so the writer knows what to keep doing
Output Format
- Structure: Voice baseline → line-by-line flags → dimension scorecard → prioritized revision list + wins
- Length: 600–1,000 words, scaling with draft length
- Format: Flags as a table or structured bullets (Original / Issue / Severity / Rewrite); scorecard as a labeled 5-item list with total and verdict
- Sections: Voice Baseline → Off-Brand Flags → Voice Scorecard → Fix First → On-Brand Wins
- Tone: Copy director in a review session — direct, specific, constructive; critiques the copy, not the writer
Quality Checklist
Before delivering output, verify:
- [ ] Every flag includes all four fields (Original, Issue, Severity, Rewrite) — no partial flags
- [ ] Each rewrite is written in the brand's actual voice, not a generic improvement
- [ ] Every flag names the specific trait, banned term, or positioning anchor it violates
- [ ] All five voice dimensions are scored 1–5 with a line-specific justification and a total out of 25
- [ ] A verdict is assigned and the Fix First list is ordered by severity
- [ ] At least 2 on-brand wins are cited by exact quote
Notes
- The most common failure is drift toward generic marketing language — vague benefit claims, hedging, and buzzwords that any competitor could have written. Weight these as Medium or High even when the grammar is fine.
- When guidelines and the draft conflict on a defensible judgment call, defer to the documented framework and note the tension rather than overriding it silently.
- For new-writer onboarding, lead with the On-Brand Wins so the review builds pattern recognition, not just correction.
How to use this skill
This file is a Claude Code skill in Anthropic's standard format — markdown with YAML frontmatter at the top.
Installing in Claude Code:
~/.claude/skills/<folder-name>/SKILL.md— auto-loads in every session anywhere on your machine (recommended for skills you'll use across multiple projects)<your-project>/.claude/skills/<folder-name>/SKILL.md— auto-loads only when working in that project (recommended for project-specific skills)
Restart Claude Code or open a new session to register.
Using in Claude.ai, ChatGPT, Gemini, or other AI tools:
- Drag this file into a new chat, or copy and paste its contents
- Send your task or request as your next message — the AI will apply the skill where it fits
For Claude Code skill details: https://docs.anthropic.com/en/docs/claude-code/skills