Brand Voice

Overview

Brand Voice loads the team's saved brand profile and applies it so every asset sounds like the team, not like generic AI. It governs voice, messaging, style, and compliance for all content the plugin produces.

The profile is a set of Markdown files the team owns in its own repository. Four files define the brand: brand-voice.md (personality, tone, signature phrases, words to avoid), messaging.md (positioning, value props, personas, boilerplate), style-guide.md (formatting, glossary, product and trademark casing, banned words, inclusive language), and compliance.md (approved claims, prohibited terms, required disclaimers). This skill is instruction-driven: it reads those files and writes on-brand copy, with no accounts, keys, or network access.

Prerequisites

• A brand profile in content/brand/ (default) or content/brands/<name>/ when a brand is named, optionally with content/brand/locales/<xx-XX>/ overrides. Create one with /brand-setup.
• Read access to the profile files. The skill uses Glob to locate them, Read to load them, and Write to save finished copy.
• No external services. All judgment comes from the profile files, never general opinion.

Instructions

1. Check for an active brand profile: Glob content/brand/, or content/brands/<name>/ when a brand is named, plus any matching locales/<xx-XX>/ directory.
2. Read all four profile files. Locale files override the base file of the same name field-by-field.
3. Set the working voice from brand-voice.md: personality adjectives, tone, do and don't rules, signature phrases, and the words-to-avoid list.
4. Apply messaging from messaging.md: pick the persona and the key message the asset reinforces, and reuse approved value props and boilerplate.
5. Configure the style layer while drafting: formatting rules, glossary terms, exact product and trademark casing, banned words, and inclusive-language rules.
6. Verify compliance against compliance.md: use only approved claims, include every required disclaimer, and remove any prohibited term.
7. Run the self-check before returning: confirm no word-to-avoid, no banned word, correct casing, and required disclaimers present. Revise until clean.
8. When no profile is found, tell the user to run /brand-setup and offer to proceed with sensible defaults for this run only.

Output

On-brand text ready to hand to the next step: a channel adapter, a localizer, or the user. When invoked by /review, the output is instead a scorecard marking Voice, Style, and Compliance as pass or fix per asset, with a redline that shows the original text, the rule it breaks, …

Channel Formats

Overview

Channel Formats adapts a single source into the shape each platform expects, then hands the draft to the brand layer. Format is the container; brand voice is the content.

Every channel has its own limits and conventions: a LinkedIn post is not an X thread, and a YouTube description is not an Instagram caption. This skill reads the spec for the requested channel from the plugin's bundled channels/<id>.md files, drafts to that spec, and enforces the length limit. It is instruction-driven and writes ready-to-paste files, with no accounts, keys, or network access.

Prerequisites

• The source content to adapt (a file path or pasted text) and one or more target channels.
• The bundled channel specs, which ship inside this skill at channels/. The skill uses Glob and Read to load the right spec and Write to save each derivative.
• An active brand profile so the brand-voice skill can run on top; if none exists, the caller is pointed to /brand-setup.

Channels (canonical IDs)

The channel ID is also the output filename (for example linkedin.md).

• linkedin — professional single post — channels/linkedin.md
• x-thread — multi-post thread — channels/x-thread.md
• newsletter — email newsletter section — channels/newsletter.md
• instagram — caption plus hashtags — channels/instagram.md
• youtube — description with timestamped chapters — channels/youtube.md
• short-video — TikTok, Reels, or Shorts script — channels/short-video.md
• blog — short post or excerpt — c…

Transcreation

Overview

Transcreation adapts content into another language so it feels native, preserving intent, brand voice, and emotional impact rather than translating literally. The wording is reworked; the effect is kept.

Literal translation breaks marketing: idioms fall flat, humor is lost, and register goes wrong. This skill loads the base brand profile and the target locale's overrides, reworks the copy for that market, protects do-not-translate glossary terms, and re-checks compliance and channel limits for the new language. It is instruction-driven and writes localized files, with no accounts, keys, or network access.

Prerequisites

• Source content to localize and one or more target locales in xx-XX form (for example de-DE, ja-JP).
• The base brand profile plus any content/brand/locales/<xx-XX>/ overrides. The skill uses Glob and Read to find and load them and Write to save each localized asset.
• The style-guide.md glossary section (Product and Trademark Names) for do-not-translate terms, and the locale's compliance.md for market-specific rules.

Instructions

1. Read the base brand files, then read any locales/<xx-XX>/ overrides. Locale rules win for that market.
2. Adapt, do not translate: rework idioms, humor, and references into locale equivalents, keeping the effect rather than the words.
3. Set the register: choose the correct formality and honorifics for the persona and locale (Sie or du, keigo, tu or usted).
4. Convert locale specifics: units, currency, date and number formats, names, and examples.
5. Protect the glossary: never translate product names, trademarks, or taglines unless the brand supplies a sanctioned translation.
6. Check the channel limit: text expands or contracts across languages, so re-fit each asset with the channel-formats spec.
7. Verify compliance against the locale's compliance.md: disclaimers and prohibited terms can differ by country.
8. Handle scripts: apply right-to-left rules for Arabic and Hebrew and correct spacing for CJK.
9. When asked, emit a back-translation plus a short note on adaptation choices so a non-native approver can sign off.

Output

One localized file per asset per locale, mirroring the source filenames under a locale folder (for example de-DE/linkedin.md). Each asset is transcreated, glossary-safe, compliance-checked for that market, and re-fitted to its channel limit. When a back-translation is requested, each localized asset is paired with a literal back-translation and an adaptation note for approver sign-off.

Error Handling