agentleFS
lauradfarbolauradfarbo/social-draft-desk/skills/brand-voice-guide/SKILL.md

brand-voice-guide

Skill4 starsChanged 24 days agoMITView on GitHub

Builds a social media voice guide for a small B2B company from its existing content, so drafting agents and reviewers write on tone. Use when asked to "build a voice guide", "define our tone for LinkedIn", "write on-tone and off-tone examples", "create a words to use and avoid list" or "set up brand voice for the content agent".

4 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

11 rules. 0 are common, 10 are distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 4 edited versions we read keep these.

  • 10

    Distinctive

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    If the social channel needs a login, ask the person to sign in.

    line 26
  2. Distinctive:

    Ask these first. Each one changes the output.

    line 30
  3. Distinctive:

    Choose three to five on-tone examples: Use the company's strongest published posts first, lightly corrected, then write one or two new …

    line 45
  4. Distinctive:

    List open items for the company to confirm: Look specifically for inconsistent product or program names across sources, unconfirmed …

    line 51
  5. Very specific:

    Use templates/voice-guide-template.md for the structure and examples/example-voice-guide.md for the expected depth.

    line 53
  6. Distinctive:

    Never invent proof points, customers, statistics or certifications.

    line 57
  7. Distinctive:

    Never put a real post in the off-tone section.

    line 58
  8. Distinctive:

    Never include pricing, funding, order volumes or internal program terms, even if a source mentions them.

    line 59
  9. Distinctive:

    Keep the guide to what a drafter needs.

    line 60
  10. Distinctive:

    Read the finished document back from its saved location and check formatting survived conversion (tables, bold labels, special characters …

    line 69
  11. Distinctive:

    Check the guide against its own rules: spelling variety, no em dashes, product names consistent throughout.

    line 70

“Common” is measured: kept by at least 6 in 10 of the 4 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Brand voice guide

Produce one voice guide that a drafting agent loads before every post and a reviewer uses as the approval checklist. The guide is built from what the company has already published, not from a blank page.

Assumptions

  • The company has some published material: a website, at least one long-form asset (guide, white paper, deck) and some social posts.
  • One channel is in scope at a time. Start with the channel the company posts on most, usually LinkedIn.
  • A person at the company will confirm the open items the guide raises.

Inputs

InputUsed for
Company overview or About textPositioning and audiences
One or two long-form assetsVoice at length, proof points, product names, approved claims
The last 10 to 15 social posts on the channelCurrent voice, best performers, drift
Founder posts reshared by the companyFounder voice, which often sets the brand voice in small teams
Any existing brand or writing rulesConstraints that override defaults

If the social channel needs a login, ask the person to sign in. Never enter credentials yourself.

Settings to confirm before drafting

Ask these first. Each one changes the output.

  1. Spelling variety. Match what the company already publishes, even when it differs from your own default. Mixed spelling across posts reads as careless.
  2. Units and date style for posts.
  3. Channels in scope.
  4. Format and location of the finished guide, so it lives where the drafting agent can read it.

Method

  1. Read everything before writing anything. Note recurring phrases, sentence length, how the company talks about competitors, how it signs off, and which posts got the most engagement.
  2. Write the positioning in one paragraph: who it is for, what it is, how it differs, and why the team is credible. Follow it with the one belief every post should reinforce.
  3. Map the audiences. For each one, what they care about and what the company leads with.
  4. Define four to six voice traits. Each trait gets a one-line meaning, a "sounds like" line and a "not like" line. Take the "sounds like" lines from the company's own best writing wherever possible.
  5. Set a starting content mix by theme, with a review point after 90 days.
  6. Write the post mechanics: opening line, length, structure, call to action, link placement, hashtags, emoji and exclamation policy, how to name founders and products.
  7. Choose three to five on-tone examples. Use the company's strongest published posts first, lightly corrected, then write one or two new drafts to cover themes the published posts miss. Explain why each one works.
  8. Write one or two off-tone examples. Invent them. Never use a real post as the bad example, because the guide gets shared. Cover the two most likely failures: generic hype, and overclaiming or attacking a competitor. List the specific reasons each fails.
  9. Add a "drift to watch" section listing patterns in recent published posts that pull away from the voice: stock openers, filler bullets, typos. Keep it factual and short.
  10. Build the words to use and avoid table with a reason for every row. Reasons are what let an agent handle words that are not on the list.
  11. Write the agent rules: human approval before publishing, a source for every statistic, no customer names without permission, no unannounced pricing, dates or terms, no criticism of named competitors, approved wording only for certifications and safety or regulatory claims.
  12. Add a pre-publish checklist a reviewer can run in under a minute.
  13. List open items for the company to confirm. Look specifically for inconsistent product or program names across sources, unconfirmed taglines, unit or spelling mismatches, and claims that need approved wording.

Use templates/voice-guide-template.md for the structure and examples/example-voice-guide.md for the expected depth.

Hard rules

  • Never invent proof points, customers, statistics or certifications. Quote only what the sources state, and name the source.
  • Never put a real post in the off-tone section.
  • Never include pricing, funding, order volumes or internal program terms, even if a source mentions them.
  • Keep the guide to what a drafter needs. If a section would not change a draft, cut it.

Outputs

  • One voice guide document in the agreed format and location, versioned and dated.
  • A short summary to the person: what the guide is built from, the drift found, and the open items that need their decision.

Verify

  • Read the finished document back from its saved location and check formatting survived conversion (tables, bold labels, special characters such as emoji in the off-tone examples).
  • Check the guide against its own rules: spelling variety, no em dashes, product names consistent throughout.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.