agentleFS
nicolas-dieznicolas-diez/personal-os/skills/build-in-public/SKILL.md

build-in-public

Skill4 starsChanged 9 days agoMITView on GitHub

Security gate and checklist before publishing anything of your own to GitHub, plus the step of telling its story. Use when the user says "let's publish this", "push to github" or "make it public".

7 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

2 rules. 0 are common, 2 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 7 edited versions we read keep these.

  • 2

    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:

    Before publishing, check how the builders you follow package similar things: naming, structure, README conventions.

    line 24
  2. Distinctive:

    Keyword check: before naming (what do people actually search?).

    line 28

“Common” is measured: kept by at least 6 in 10 of the 7 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

build-in-public

Meta-story: this is the skill that published itself. It started as "publish to github" (the mechanical step) and was renamed to what the step is actually for — building in public. The pattern comes from watching how effective builders ship: build → open source → tell the story. The README is half the positioning value.

Layer: static checklist, NOT an agent. Scope: manual work before publishing. Automate only after you've published 2–3 repos by hand.

0. Security gate — FIRST, and blocking

Mandatory. If anything fails or is ambiguous, STOP and ask. Don't publish.

  • Scan every candidate file for personal/private data: real emails, personal addresses, absolute home paths (/Users/...), family names, confidential employer or hiring-process material, credentials, compensation, government IDs.
  • Also scan for: (a) numbers or findings from take-home tests of active hiring processes (an employer can google the candidate and find their case dissected); (b) internal numbers from ex-employers (a generic framework publishes; internal figures don't); (c) personal-context lines that undermine positioning (valid in a private version, never in the public one); (d) indirect identifiers a grep won't catch: exact parameters of a brief or case (formats requested, amounts, dates, specific thresholds) — generalize or round them, then re-read asking "could someone who knows the case map this back?".
  • What gets published is a sanitized adaptation, not the original. Generic placeholders are fine. Doubt wins: stop and ask.
  • Suggested grep (adapt to your own data): your-email|your-last-name|/Users/|@kindle|gmail\.com|<employer>|<active-process-companies>|token|password|interview.

1. Survey the referents

Before publishing, check how the builders you follow package similar things: naming, structure, README conventions. Reuse what converges. (This is your learning copilot's survey mode.)

2. Marketing / positioning research (GitHub)

  • Keyword check before naming (what do people actually search?). Language: English by default for the public version (discoverability), unless you explicitly decide otherwise.
  • Topics = discoverability driver #1: utilitarian + positioning (e.g. data-analysis, claude-skill, ai-native, product-management).
  • README: keyword in the H1 and first paragraph; repo description carries the keyword.

3. Frame it in your thesis

First paragraph of the README = why you built this (your point of view; e.g. AI-native = building your own tools; feedback + data → synthesis → prototype; evals-first). Link to your flagship project and profile. Credit public frameworks and influences; keep private material out.

4. Structure and push

Standalone repo per tool, OR a folder in an existing repo if you explicitly prefer. README.md + LICENSE + .gitignore (+ SKILL.md if it's a skill). Manual push the first 2–3 times — the checklist leaves everything ready; you push.

5. Tell the story WITHOUT an audience

You don't need to become a poster to build in public. A README-first, 1:1-reach version works:

  • The README is the story: the real problem that led to building it → what it does in one sentence → the most interesting methodological detail → how to install it. Written for someone arriving via a direct link, not a feed.
  • One presentation line ready to paste in a DM, an email, an application, or to say in an interview: "I built X to solve Y — here's the repo." Two variants: casual (DM) and formal (application/interview).
  • Profile up to date: the profile README ties the repos together; it's your landing page.
  • Social post: ONLY if you decide it that day. This checklist never requires one.

Quick checklist

  • Security gate passed (including active take-homes and ex-employer internals)
  • Referents surveyed
  • Keywords + topics + language decided
  • README framed in your thesis + links
  • LICENSE + .gitignore ready
  • Share kit delivered (presentation line ×2 + README as the story)
  • Manual push done by you

Output

Sanitized artifact ready to push + suggested topics + description + 1:1 share kit. Nothing pushes itself.

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.