agentleFS
prateeks367prateeks367/voicemoat-skills/skills/write-in-my-voice/SKILL.md

write-in-my-voice

Skill0 starsChanged 42 days agoMITView on GitHub

Draft a Twitter or LinkedIn post that sounds like the person asking, by working out their voice from writing they have already published and checking the draft against it. Use when someone asks for a post, a thread, a hook or a rewrite, or asks whether a piece of text sounds like them. Do not use it for writing that is not going on social media.

3 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

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

  • 0

    Common

    Most of the 3 edited versions we read keep these.

  • 10

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Never skip to drafting. A post written before you know how this person writes is a guess, and it will read like one.

    line 13
  2. Distinctive:

    Ask for five recent posts they wrote themselves, pasted in.

    line 18
  3. Distinctive:

    If they cannot paste posts, ask these four instead and work from the answers:

    line 21
  4. Distinctive:

    Do you open with a claim, a story, or a question?

    line 24
  5. Distinctive:

    If they give you neither, say plainly that you can write a decent generic post but not one in their voice, and let them choose.

    line 28
  6. Distinctive:

    Write the post yourself, against the sketch.

    line 53
  7. Distinctive:

    If they are stuck on an opening, offer three from genuinely different angles: contrarian, story, and reframe.

    line 56
  8. Distinctive:

    Before you hand the draft over, go through the sketch line by line and say what matches and what does not.

    line 61
  9. Distinctive:

    Never claim a numeric score.

    line 70
  10. Distinctive:

    Never make the second call on your own.

    line 98

“Common” is measured: kept by at least 6 in 10 of the 3 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

Write it in my voice

Most AI drafts fail the same way. They are competent and they sound like nobody. The fix is not a better prompt, it is having something concrete to match against, so this workflow builds that first and then measures the draft against it.

Never skip to drafting. A post written before you know how this person writes is a guess, and it will read like one.

Step 1: get something real to match against

Ask for five recent posts they wrote themselves, pasted in. Not their favourite posts, and not posts they admire from other people. Ordinary ones.

If they cannot paste posts, ask these four instead and work from the answers:

  1. What do you write about, and who reads it?
  2. Do you open with a claim, a story, or a question?
  3. Short and punchy, or a longer build?
  4. What would you never say? Words, phrases, or a whole register.

If they give you neither, say plainly that you can write a decent generic post but not one in their voice, and let them choose.

Step 2: build the sketch, and show it to them

From what you have, write down:

  • Openings. How the first line earns the second. Claim, story, number, question, contradiction.
  • Length and shape. Sentence length, paragraph length, whether they use line breaks as punctuation.
  • Vocabulary. Words they reach for. Words they clearly avoid.
  • Register. Formal, blunt, warm, dry, self-deprecating.
  • Never. Emoji, hashtags, hype words, engagement bait, exclamation marks. Whatever the samples show they do not do.

Show this sketch to the person before you draft, in about six lines. They will correct one thing, and that correction is worth more than another three samples.

‼️ Be honest about what it is. A sketch from five posts is a sketch from five posts. Do not call it their voice profile and do not imply you have analysed their whole history.

Step 3: draft

Write the post yourself, against the sketch. One draft, not three, unless they asked for options.

If they are stuck on an opening, offer three from genuinely different angles: contrarian, story, and reframe. Let them pick rather than picking for them.

Step 4: check it, out loud

Before you hand the draft over, go through the sketch line by line and say what matches and what does not. Something like: the opening is a claim, which fits; the third sentence is longer than anything in your samples; "leverage" does not appear once in five posts and it is in this draft twice.

Then fix what you named. Once. If it is still off after one pass, say what you cannot get right and ask them what you are missing, rather than rewriting the same post a fourth time and hoping.

Never claim a numeric score. You are comparing against a handful of samples, not measuring, and a number would imply a precision you do not have.

Step 5: hand it over honestly

Give them the post, and one line on what you were unsure about. A draft delivered with its weak spot named gets edited. A draft delivered with a flourish gets posted and regretted.

What changes when VoiceMoat is connected

Everything above works with nothing installed. With a VoiceMoat account connected, the guesswork in steps 1, 2 and 4 is replaced with real data:

  • get_voice_profile returns a voice profile trained on your own published posts rather than a sketch from five pasted posts, including the phrases you never use.
  • score_voice_match checks the draft against that profile in two parts: Style, how many of your measurable habits it matches (N of M checks), and Stance, whether what it says fits your views (On voice, Check one line or Off voice), with the lines that are off. The assistant may show it as a card. That is a measurement, not the read-through in step 4.
  • get_voice_insights returns your dominant patterns, hook style and sentence rhythm, and how far your recent posts have drifted from your own baseline.
  • publish_post and schedule_post put the finished post on Twitter or LinkedIn, behind a preview and a one-time confirmation so nothing goes out that you have not read.

Never make the second call on your own. Show the person the exact preview the first call returns and wait for a clear yes before you send the one-time code back. Never make both calls in one step. The assistant may show the preview as a card with its own Post now or Schedule button. A press on that button is the person's yes and sends the code itself, so do not send it again.

VoiceMoat is at voicemoat.com. The connector needs the Pro or Enterprise plan.

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.