Brand Voice Enforcer

Reviews any draft against your brand voice document — returning an overall rating, specific flags by severity with quoted passages, suggested rewrites for every notable issue, and coaching notes the writer can apply to future work.

What it does

Produces a structured voice compliance assessment across six sections:

• Overall assessment — one paragraph covering whether the draft broadly sounds like the brand, its strongest alignment point, its primary concern, and an overall rating: STRONG / GOOD / NEEDS WORK / SIGNIFICANT REVISION NEEDED
• Voice strengths — at least two specific passages that align well, each quoted, explained, and linked to the relevant voice principle
• Voice flags — each meaningful departure quoted, with the conflicting principle named, the exact issue explained, and a severity rating: MINOR / NOTABLE / SIGNIFICANT. Stylistic variation within the acceptable brand range is not flagged.
• Vocabulary flags — banned words or avoided phrases used, missed brand phrase opportunities, and terminology inconsistency
• Suggested rewrites — for every NOTABLE or SIGNIFICANT flag: [Original] → [Revised] with a one-line explanation
• Writer notes — 2–3 constructive observations the writer can apply to future work, focused on recurring strengths and recurring risks

If no brand voice document, approved writing sample, or explicit voice principles are provided, the skill stops and requests them before proceeding — it will not infer a voice standard from generic marketing best practices.

When to use it

Use it as the final QA step before publishing or sending any customer-facing copy. Works on emails, blog posts, landing pages, social copy, and ads. Best deployed in a Claude.ai Project with the brand voice document attached as project knowledge so it's always available. Use Sonnet for routine checks; Opus for longer content or nuanced assessment. The output from any of the other skills in this stack feeds directly into it.

When not to use it

• Without a brand voice document — the skill requires a reference standard and will not proceed without one
• As a substitute for editorial judgement — it flags rule-breaking against the provided framework, not poor writing in general
• For generating copy from scratch — use the other skills in this stack for drafting, then bring this in at the end