# Brand Voice Codex > Activate whenever a brand's voice needs to be defined, documented, enforced, or repaired — creating a voice guide, keeping copy consistent across channels and writers, onboarding collaborators or AI tools to write in-voice, diagnosing "this doesn't sound like us," or managing multiple sub-brands under one parent identity. Trigger on any voice, tone, or brand-language work. Fire because unwritten voice lives in one person's head, drifts with every new writer and channel, and dies at scale — a brand that cannot state its voice cannot delegate its words. - Skill: `forexgod21/brand-voice-codex` (Agent Skill, multi-file: 2 files) - Install (CLI): `npx skillmds@latest add forexgod21/brand-voice-codex` - Raw SKILL.md: https://api.skillmd.com/api/skills/forexgod21/brand-voice-codex/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Marketing & Growth - License: CC BY 4.0 - Author: Forexgod21 (https://skillmd.com/u/forexgod21) - Updated: 2026-09-21 - Page: https://skillmd.com/skills/forexgod21/brand-voice-codex --- # Brand Voice Codex **Attribution:** YourVisionYourCreation LLC — yourvisionyourcreation.com **Doctrine class:** YVYC original — marketing category --- ## Universal So-What Voice is the brand's fingerprint — the thing that makes a reader know who is talking before they see the logo. Unwritten, it exists only as the founder's instinct, and instinct does not transfer: every new writer, channel, and AI tool approximates it a little differently until the brand sounds like a committee. The codex converts instinct into doctrine — written, enforceable, and transmittable — so the voice survives delegation and scale. --- ## Core Doctrine ### 1. Voice Is Character, Tone Is Weather The codex's foundational separation: - **Voice** is WHO the brand is — stable across every context: its values, its stance, its relationship to the reader. Voice does not change. - **Tone** is HOW the voice sounds in a given moment — modulating by context: celebratory at a launch, steady in an apology, plain in a service notice. Tone flexes; voice holds. - Most "voice problems" are actually undefined tone ranges: the brand knows who it is but nobody wrote down how it behaves at a customer complaint versus a product drop ### 2. The Codex Core — Five Definitions The written voice document contains, minimum: | Element | Contents | |---|---| | **Character** | The brand-as-person: three to five trait words, each with a sentence of what it means HERE — "bold" defined, because every brand claims bold | | **Stance** | The relationship to the reader: guide, peer, expert, coach, co-conspirator — one primary stance, held | | **This-not-that pairs** | The traits fenced on both sides: "confident, not arrogant; warm, not saccharine; direct, not blunt" — the NOT side is where drift happens | | **Lexicon** | Words the brand uses, words it never uses, how it names its products, its customers, and itself — with the banned list explicit | | **Signature moves** | The recognizable habits: sentence rhythms, how it opens, how it closes, its relationship to humor, emoji, jargon, capitalization | ### 3. Canonical Examples Carry the Codex - Every definition is anchored by REAL examples: three to five canonical passages of the brand at its best, annotated — "this is the voice, and here is why" - Anti-examples are equally load-bearing: passages that are almost right but off, annotated with what broke — the near-miss teaches more than the bullseye - Rewrite demonstrations close the loop: the same message shown off-voice and on-voice, side by side — the transformation IS the training ### 4. The Tone Map - The codex maps tone across the contexts the brand actually meets: launch, sale, apology, support reply, crisis, social banter, legal-adjacent notices — each with its register and its example - The hard contexts get the most detail: anyone can sound like the brand at a product launch; the codex earns its existence at the refund dispute and the public mistake - The invariant rule: whatever the tone, the CHARACTER holds — an apology in-voice is still recognizably the brand, not a lawyer wearing its logo ### 5. Sub-Brand Architecture For families of brands under one parent: - Each sub-brand states its inheritance: what it takes from the parent (values, stance, standards) and what it expresses in its own register (audience, energy, lexicon) — inheritance explicit, never assumed - The separation is enforced in both directions: parent-register language does not leak into a sub-brand built for a different audience, and a sub-brand's register does not migrate up — register contamination kills both identities - One codex owner per brand; the parent codex owner arbitrates cross-brand conflicts ### 6. Enforcement and Drift Watch - The codex is the onboarding package for every writer, agency, and AI tool BEFORE they produce — correcting off-voice work after publication costs reach and trust - The voice review is a named step in the publishing path: work is checked against the codex (character, this-not-that, lexicon, signature moves) before it ships - Drift audit on a cadence: sample published work across channels, score against the codex, and trace drift to its source — a channel, a writer, a tool — then fix the source, not just the samples - The codex is versioned: voice evolves deliberately, by decision and record, never by accumulation of small unreviewed drifts --- ## Common Failure Modes | Failure | Cause | Correction | |---|---|---| | Brand sounds different on every channel | Voice lives in one head | The written codex, five definitions minimum | | "Bold" produces obnoxious | Trait words undefined and unfenced | This-not-that pairs; anti-examples | | Apologies sound like legal filings | No tone map for hard contexts | Hard contexts get the most codex detail | | New writer needs months to sound right | No canonical examples | Annotated examples + rewrite demonstrations | | Sub-brand bleeds into the parent | Inheritance never made explicit | Sub-brand architecture, both-direction enforcement | | Voice erodes one post at a time | No drift watch | Cadenced audit; fix the source of drift | --- ## Non-Negotiables 1. The voice is written, or it does not exist at scale. 2. Voice holds; tone flexes — never confused. 3. Every trait is fenced with its this-not-that pair. 4. Canonical examples and anti-examples anchor every definition. 5. Sub-brand inheritance is explicit in both directions. 6. Voice changes by decision and record, never by drift. --- *YourVisionYourCreation LLC — yourvisionyourcreation.com* *Licensed under CC BY 4.0*