agentleFS
Fadi88Fadi88/one-man-startup/.claude/skills/founder-voice/SKILL.md

founder-voice

Skill16 starsChanged 18 days agoMITView on GitHub

How to word anything the founder reads or answers - questions, grilling rounds, approvals, decisions, summaries, reports - at the technical level saved in project.toml [owner] level (non-technical, basics or technical). Use before writing any message or document for the founder.

2 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

4 rules. About 0 are common, 3 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    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:

    When more than one question is open, number them and ask for answers by number ("1 yes, 2 no").

    line 29
  2. Distinctive:

    Use the plain version, or say the thing it does:

    line 41
  3. Distinctive:

    When there is no plain word, explain the idea in one sentence and drop the term.

    line 57
  4. Very specific:

    Every decision the founder is asked for, in chat or in p1-open-questions.md, is one card: decision, recommendation, what could go wrong, …

    line 64

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “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
<!-- Generated by tools/sync_adapters.py from skills/founder-voice/SKILL.md. Edit the source, not this file. -->

Founder voice

The founder's level is in project.toml [owner] level. Read it before you write to them. If it is missing or unset, the first question you ask is the level itself (below), worded for a non-technical reader. Never guess it from how they write.

The three levels

LevelWhoHow you write
non-technicalHas never written code.No technical words at all. Every choice as what it means for the business, what it costs, what could go wrong, and your recommendation. Say what the founder will see or feel, not how it works.
basicsKnows the ideas (a database, a server, a deploy) but does not code.Plain words first; a technical term is fine when it is the clearest one, explained once in a few words the first time. No code, no commands, no file paths unless they asked.
technicalWrites or reads code.Brief. Technical terms, file paths and diffs are fine; lead with the decision and the diff.

Asking for the level (setup, or the first grilling when it is unset): "How technical are you?

  1. I've never written code. 2. I know the basic ideas but don't code. 3. I write or read code. My suggestion is 1 unless you're sure; you can change it any time."

Every message, at every level

  • One decision per question, with your recommendation, worded so that "yes" accepts it.
  • When more than one question is open, number them and ask for answers by number ("1 yes, 2 no"). A bare "yes" with several open is asked back, never read as covering all of them; accepting every recommendation is said outright ("accept all") or marked in the choices file.
  • Screens and pictures are shown, never pointed at: in the chat where the tool can show an image, or as one generated page that opens by itself. Never a file path to open.
  • Facts you can look up, you look up; the founder is asked only for decisions.
  • Numbers keep their units and their source: "about 20 dollars a month (provider's price page)".
  • The founder can change their level any time ("explain less", "I don't follow"). Update project.toml [owner] level the same day, with the date in p1-closed-decisions.md.

Words for non-technical

Use the plain version, or say the thing it does:

Instead ofSay
deploy, release to productionput the new version live for your users
rollbackput the previous version back
repository, repoyour project's folder on GitHub
commit, push, mergesave the change, send it to GitHub, add it to the live version
database, migrationwhere the data is kept; a change to how it is stored
API, endpoint, integrationa connection to another service
frontend, backendthe part users see; the part behind it
bug, regressionsomething broken; something that used to work and now doesn't
test, CI, sweepan automatic check; the checks that run after every change
requirement row, specone sentence the product must do; the written plan for a feature
environment variable, secret, API keya private setting; a password for a service

When there is no plain word, explain the idea in one sentence and drop the term.

Approvals

Say what will happen, what it changes for users or money, whether it can be undone and how long that takes, then ask.

Every decision the founder is asked for, in chat or in p1-open-questions.md, is one card: decision, recommendation, what could go wrong, how to undo it, and two replies to copy, the recommended one first (TPL-113). The shape, an example and the wording at each level are in skills/founder-voice/card.md.

Saying something is done

A report that something is done, fixed or live ends with one thing the founder can check themselves in under a minute, without reading code (TPL-116): "Open yourapp.com/signup on your phone: the new page has the green button." Or: "Open the sheet: row 14 has today's orders." A report with nothing the founder can check yet says what it is waiting on before it can be seen.

Documents the founder reads

Specs, summaries, QA and legal reports, the dashboard, release notes: write the parts the founder must read or decide on at their level, and put the technical detail other agents need in a separate section headed for them.

A review reaches the founder as its "For the founder" list (agents/reviewer.md), word for word, with how many lines block launch and how many can wait; never summarised into "a few issues" (TPL-106).

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.