Brand Voice Guardian
So AI sounds like your brand — not like AI.
SKILL.md in the open agentskills.io standard — works directly in Claude, ChatGPT/Codex, Cursor, Copilot & more.
Brand Voice Guardian is an AI skill that learns, documents, applies, reviews and protects your brand's voice — so AI content sounds like your brand, not like generic AI. It builds a persistent brand-voice.profile.md, reviews new content for brand fit, tone, terminology, channel and AI-generic risk, and rewrites copy on-brand. Not a rewrite tool: a brand-governance layer for AI workflows.
What this skill does
- Learns a reusable brand voice (Voice DNA) from good/bad example texts and documents it as brand-voice.profile.md
- Cleanly separates voice (stable) / tone (situational, four measurable dimensions) / channel-style (implementation)
- Reviews text for brand fit, tone fit, channel fit, terminology, AI-generic risk, overclaim and meaning-preservation
- Rewrites on-brand with variants (clean, bolder, shorter, channel-optimized, premium, community, support-safe, legal-safe) under Meaning Lock
- Adapts tone per channel (website, LinkedIn, support, Discord, ads, docs, UI microcopy…) and per situation (onboarding, error, crisis, launch…)
- Guards terminology (product names, spellings, do/avoid words) and detects generic AI phrasing
- Protects the profile: temporary override ≠ permanent change; permanent changes only with confirmation + change log
- Exports the voice for ChatGPT (project/Custom GPT), Claude (project/skill), Cursor (.mdc), AGENTS.md and any LLM
Description
Brand Voice Guardian is a skill for brand-voice governance in AI workflows: it learns a brand voice, documents it as a reusable profile, applies it, reviews text against it and protects it from accidental drift. It follows the evidence-based three-layer model: **voice** is the stable brand personality (constant everywhere), **tone** is the situational adaptation — measured on four dimensions (funny–serious, formal–casual, respectful–irreverent, enthusiastic–matter-of-fact) rather than vague adjectives — and **channel/style** is the concrete implementation per medium. The core idea: ordinary AI writes "well"; Brand Voice Guardian writes so it sounds like *this* brand. It learns a Brand Voice DNA (personality, stance, rhythm, word choice, closeness, boundaries) from good and bad example texts, produces a **brand-voice.profile.md** with core voice, tone sliders, do/don't rules, terminology, channel rules and a situation matrix, and makes the voice usable across platforms (ChatGPT projects/Custom GPT, Claude projects/skills, Cursor rules, AGENTS.md, any LLM). A core product rule protects the brand: the profile is the persistent memory, a one-off tone request ("make this cheekier") is only a temporary override, and **permanent profile changes always require explicit confirmation** (with a change summary and change log). Built to be responsible: it never impersonates other brands or real people, never writes **fake reviews** (e.g. the FTC rule against AI-generated fake reviews), never strengthens unproven or absolute claims, makes no bare "green/eco" claims, and never alters legal or mandatory wording — it flags those for review. It uses only example texts you have rights to, treats them as confidential, and minimizes data. And it's honest about persistence: it doesn't "remember" on its own — it produces the profile and the exports you store in each platform.
Examples
What it does not do
- Doesn't impersonate real, identifiable brands or a real person's voice for commercial use
- Doesn't write fake reviews/testimonials/endorsements or fake followers; no astroturfing
- Doesn't strengthen unproven or absolute claims and makes no bare "green/eco" claims; invents no facts
- Doesn't alter, shorten or omit legal, medical, financial or mandatory wording — flags it for review
- Doesn't change the saved profile from a one-off request; doesn't overwrite/delete a profile without confirmation
- Doesn't pretend to "remember" — persistence lives in the platform where you save the profile/export
Compatibility & tech
- Tested (internal)
- 5 scenarios
- Modes
- Train Brand Voice (setup) · Activate / Deactivate (state + scope) · Voice Fit Review · Brand Rewrite (Meaning Lock) · Channel Voice Adapter · Tone Situation Matrix · Terminology Guardian · AI-Generic Risk Check · Voice Drift Check · Campaign Consistency Check · Localization Voice Adapter · Brand Risk Review (overclaim/greenwashing/sensitive statements) · Edit Profile (confirmation required) · Export (ChatGPT/Claude/Cursor/AGENTS.md/generic)
- Inputs
- Brand, offer, audience, channel · good and bad example texts (only your own/licensed) · a text to review or rewrite · an existing brand-voice.profile.md · terminology / no-go lists · channel + situation
- Output format
- Quick Voice Check (score + top fixes), Detailed Brand Voice Report, Rewrite Pack (original → on-brand → variants → "what changed"), Change Proposal (profile edits only), and exports (brand-voice.profile.md + platform formats). Native tables/checklists/buttons where available, else A/B/C text fallback.
- Subcategory
- Brand voice & content governance
- License
- Proprietär
Security profile
Runs entirely on your machine with your own AI — no external runtime, no running costs.
Only instructions, templates and references — no executable scripts.
Works offline with what you provide — does not call external services on its own.
What you get
- brand-voice-guardian-1.0.0/12 files
- .claude-plugin/marketplace.json
plugins/brand-voice-guardian/skills/brand-voice-guardian/12 files
- SKILL.md
- manifest.json
references/10 files
- brand-voice-guardian-checklist.md
- brand-voice-method.md
- brand-voice-profile-template.md
- change-protection.md
- channels-and-situations.md
- example-profile.md
- platform-exports.md
- review-and-rewrite.md
- safety-and-rights.md
- terminology-and-ai-generic.md
- .agents/skills/brand-voice-guardian/→ universal — same content (Codex, Cursor, Copilot, Gemini, Windsurf, Cline)
- LICENSE.txt
Installation
Also works as a chat prompt
No AI tool? Paste it into Claude, ChatGPT or Gemini and use the method right away.
This skill ships no scripts, so the prompt carries its full method.
Installing is the full version — it triggers automatically, runs its scripts and loads references as needed. As a chat prompt you drive the method by hand.
Unlock to copy the ready-to-paste prompt — then in “My Skills”.
Reviews
No reviews yet – be the first.
Note
A tool for your own, authorized brand voice — not legal advice. Brand Voice Guardian doesn't impersonate other brands/people, doesn't write fake reviews and doesn't strengthen unproven claims; it leaves legal, medical, financial and mandatory wording unchanged and flags it for professional/legal review. Use example texts only if you have the rights; treat them as confidential and minimize data. Persistence lives in the platform you use.
Changelog
- v1.0.001.07.2026[object Object]
Frequently asked questions
What does Brand Voice Guardian do?
Brand Voice Guardian is an AI skill that learns, documents, applies, reviews and protects your brand's voice — so AI content sounds like your brand, not like generic AI. It builds a persistent brand-voice.profile.md, reviews new content for brand fit, tone, terminology, channel and AI-generic risk, and rewrites copy on-brand. Not a rewrite tool: a brand-governance layer for AI workflows.
How does Brand Voice Guardian make AI content sound like my brand?
Brand Voice Guardian learns a reusable brand voice from your good and bad example texts and writes it into a brand-voice.profile.md — core voice, tone sliders, do/don't, terminology and channel rules. It then reviews new copy for brand fit and rewrites it on-brand, without changing your facts or claims.
What's the difference between a permanent brand voice and a one-off tone request?
That's the core rule of Brand Voice Guardian: the Brand Voice Profile is the persistent memory. A request like 'make this one cheekier' applies only to that task (a temporary override). Permanent changes to the profile happen only after explicit confirmation — with a change summary and a change log.
Can I use the brand voice in ChatGPT, Claude and Cursor?
Yes. Brand Voice Guardian exports the voice as brand-voice.profile.md plus ready formats: ChatGPT project instructions, Custom GPT, Claude skill/project, a Cursor rule (.mdc) and AGENTS.md. Honest caveat: persistence lives in the platform (project files/repo) — the skill doesn't 'remember' on its own; it produces the file you store there.
Does the skill write customer reviews or strengthen marketing claims?
No. Brand Voice Guardian writes no fake reviews and never strengthens unproven or absolute claims (no 'clinically proven', '#1', 'guaranteed'), doesn't impersonate other brands or real people, and doesn't alter legal or mandatory wording. It flags such passages for professional/legal review.
Which example texts may I feed it for training?
Only texts you own or are licensed to use; Brand Voice Guardian won't imitate third-party copyrighted texts. It treats your brand/customer texts as confidential and works with data minimization (no passwords, credentials or unnecessary personal data).
Which AI tools does Brand Voice Guardian work with?
Claude (Projects, Skills & system prompt) · ChatGPT (Projects, Custom GPT & Custom Instructions) · Cursor (.cursor/rules/*.mdc & AGENTS.md) · Any LLM (system prompt / knowledge base) · Native UI (buttons/tables/checklists) where available, Markdown fallback otherwise
How do I use Brand Voice Guardian?
Brand Voice Guardian is a SKILL.md in the open agentskills.io standard: install it with one command (npx) or download it and add it to your AI tool — Claude (Projects), ChatGPT (Custom GPT), Cursor, Copilot, Gemini CLI and more. No code needed.