Guide · Agent skill
How to use the Brand Voice agent skill (brand-voice)
Updated · written with AI from the project's own documentation and checked against our catalogue data
The brand-voice skill is a set of instructions that teaches an AI agent to study writing you already have, such as posts, essays, launch notes, emails that worked and product docs, and to summarise its style into a short reusable profile. This guide is for people who have never installed an agent skill before. By the end you will know what the skill does, how to add it, how to trigger it with plain requests, and what to check before using it.
Brand Voiceaffaan-m/ECC · MIT Download the skill Quick start
- 1Download and unzip. Download the brand-voice ZIP from this page and unzip it.
- 2Place the skill folder. Put it in ~/.claude/skills/ for personal use or in the project's .claude/skills/ folder, or upload it under Settings → Capabilities → Skills in the Claude apps.
- 3Gather samples. Collect 5 to 20 representative posts, essays, launch notes, emails or docs written by you.
- 4Ask for a profile. Ask the agent in plain language to build a voice profile from your samples.
- 5Check the profile. Read the VOICE PROFILE block and confirm the rules match what your writing actually does.
- 6Draft with the profile. Ask for posts, emails or launch copy and review the output against the banned moves.
What the brand-voice skill makes the agent do
Without a skill, an AI agent usually guesses a style each time you ask it to write something, and the result often sounds generic. The brand-voice skill changes that. It tells the agent to build a durable voice profile from real source material and then reuse that profile for content, outreach and social posts.
The profile covers rhythm and sentence length, compression versus explanation, capitalisation, how often parentheticals, numbers and questions appear, how sharply claims are made, how transitions work, and what the author never does. The skill's output is a block called VOICE PROFILE that other skills can read directly.
- Source priority: recent original X posts and threads first, then articles, essays and launch notes, then emails or DMs that worked, then docs, changelogs and site copy.
- Generic platform exemplars are not used as source material.
- A default ECC/Affaan voice is included for when your own samples are thin: direct, compressed, concrete.
- A list of hard bans is applied to drafts, such as fake curiosity hooks, bait questions, forced lowercase and 'Excited to share'.
Install the brand-voice skill
Download the ZIP from our page and unzip it. You can install the skill in one of two places. For a personal setup, unzip it into ~/.claude/skills/. For a single project, unzip it into the project's .claude/skills/ folder.
If you use the Claude apps, you can instead upload the skill under Settings → Capabilities → Skills.
- Personal install: ~/.claude/skills/
- Project install: the project's .claude/skills/ folder
- Claude apps: Settings → Capabilities → Skills, then upload
Give the skill sample writing
The skill needs real writing to learn from. The source asks for 5 to 20 representative samples when available. Good samples are posts you wrote, essays, memos, launch notes, emails that got replies, and product docs or README text. Prefer recent writing unless you say an older piece is more canonical.
Paste or attach the samples in your request. If you have live X access through the separate x-api skill, the agent can pull your recent original posts before drafting.
- Use your own writing, not generic examples.
- Mix public and private writing only if you want the skill to separate a public launch voice from a working voice.
- Skip sample material you do not want the profile to copy.
Trigger the skill with natural requests
You do not need special syntax. Describe the job in plain language. The skill activates when you want content or outreach in a specific voice, when you write for X, LinkedIn, email, launch posts, threads or product updates, when you want to adapt an author's tone across channels, or when a content lane needs a reusable style system.
Once a VOICE PROFILE exists in a session, the agent reuses the latest confirmed profile for related tasks. If you want to keep it, ask for it to be saved to a workspace location or memory surface you name.
- Build a voice profile from these five posts and save it as a reusable block.
- Draft three launch posts for X in the same voice as my existing essays.
- Write a cold email to a potential user using our voice profile.
- Rewrite this changelog intro so it sounds like the rest of our docs, and remove any hype.
Examples of what a good profile looks like
A profile is a short structured note, not an essay. The schema lists fields such as Author, Goal, Confidence, Source Set, Rhythm, Compression, Capitalization, Parentheticals, Question Use, Claim Style, Preferred Moves, Banned Moves, CTA Rules and Channel Notes for X, LinkedIn and Email.
Every banned move should be observable in your source set or explicitly requested by you. If your samples do not support a rule, the profile should say so rather than invent one.
- Question Use: rare, used directly rather than as bait.
- Claim Style: specifics, mechanisms and numbers beat adjectives.
- Banned Moves: only patterns the source actually avoids.
Safety, scripts and network notes
The skill does not run scripts. It does use the network, but only if you enable the optional x-api skill and X API access to pull recent posts. Without that, it works from the text you supply.
The skill is designed not to create repo-tracked files that store personal voice fingerprints unless you explicitly ask. Use it to shape drafts; it does not fetch or post content on its own.
- Runs scripts: no
- Uses network: yes, optionally via x-api for recent X posts
- Does not post content automatically
- Does not commit personal voice profiles to a repo unless you ask
Tips for better results
The quality of the profile depends on the quality of your samples. Five strong, recent pieces usually beat twenty old, mixed ones. If the profile looks like guesswork, add more real writing before drafting.
Review each draft against the banned list. The skill itself says to delete and rewrite fake hooks, bait questions and generic founder-journey filler.
- Start with recent samples and label which are public and which are private.
- Re-run the profile when your writing changes.
- Check the Confidence field in the profile before relying on it.
What people use it for
- Founders keeping many posts and emails sounding like one author.
- Open-source maintainers keeping a project's public voice consistent across X, LinkedIn and launch notes.
- Marketing teams with a body of good writing who want a repeatable style guide for AI-drafted copy.
- Preparing drafts for content-engine, crosspost or outbound workflows that should share one voice.
Frequently asked questions
Is the brand-voice skill free to use?
The skill is published under the MIT licence, as stated in its SKILL.md. Check the licence file in the ZIP for the full terms. Optional live X access depends on the separate x-api skill and your own X API access.
What do I need to install brand-voice?
You need the skill folder, which you can download as a ZIP from this page and unzip into ~/.claude/skills/ or a project's .claude/skills/. You also need sample writing, such as posts, essays or emails, for the skill to learn from.
Does brand-voice run scripts or post to X for me?
No. The skill does not run scripts and does not post content automatically. It shapes drafts, and it can pull recent posts only if you add the optional x-api skill with X API access.
How is brand-voice different from just asking an AI to match my style?
Asking for a style often gives a generic guess. This skill builds a structured profile from your real writing, records rules such as sentence length and banned phrases, and reuses that profile across tasks in the same session.
Can I use brand-voice to imitate someone else's writing?
Only with that person's consent. The skill is built to mirror a specific author's writing from source samples, so it is not intended for imitating a person without permission.
Why does the profile say my voice is uncertain?
The profile includes a Confidence field. If you supply very little real writing, the profile will be closer to guesswork, and the skill prefers real sources over generic exemplars. Add more of your own samples to improve it.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
Use brand-voice if you have a body of your own writing and want AI-drafted posts, emails and launch copy to sound consistent and avoid generic AI phrasing. Pick something else if you have very little writing to learn from, want the tool to post or fetch content automatically, or need to imitate a person who has not agreed to it.
Download the skill