readme-craft
Skill4 starsChanged 18 days agoMITView on GitHub
README writing and restyling — scaffold a new README, restyle an existing one, or add/refresh its badges (shieldcn.dev or shields.io), header, admonitions, collapsibles, and GitHub markdown patterns.
7 other projects use a version of this.
Adopted by
Most recently taken up by
- guilhem/skillstheir own version, first seen Nov 1, 2026
- hanzhangzzz/agent-skills-zhtheir own version, first seen Oct 29, 2026
- uwakeme/wake-skillstheir own version, first seen Oct 25, 2026
- pyros-projects/Trialtheir own version, first seen Sep 7, 2026
- okht/readme-crafttheir own version, first seen Jul 12, 2026
- Xipher-Labs/walter-ostheir own version, first seen May 21, 2026
Other versions7 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 7
- Xipher-Labs/walter-os★ 8same name
- guilhem/skills★ 2same name
- pyros-projects/Trial★ 2same name
- hanzhangzzz/agent-skills-zh★ 1same name
- motiful/readme-craft★ 1same name
- uwakeme/wake-skills★ 1same name
- okht/readme-craft★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository17 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- ask-firstSkill
- claude-project-rulesSkill
- compact-markdownSkill
- humanizeSkill
- pr-reviewSkill
- empty-commitSkill
- github-issue-creatorSkill
- git-mergeSkill
- git-worktreeSkill
- handoff-sessionSkill
- html-landing-pageSkill
- offload-taskSkill
- skill-auditSkill
- study-notesSkill
- visualize-itSkill
- zoom-outSkill
The file
We split this skill into the rules it gives an agent
6 rules. 0 are common, 3 are distinctive, and 3 are very specific to this project.
0
Common
Most of the 5 edited versions we read keep these.
3
Distinctive
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 12
Before writing or touching any badge, ask via AskUserQuestion: shieldcn.dev or shields.io.
- Very specific: line 38
Every badge: live or meaningful signal + <a href> to the page it represents (stargazers, npm page, workflow runs, LICENSE, funding page).
- Distinctive: line 66
No dark/light wrapping:
- Distinctive: line 100
Ask before writing if any of these are unclear — don't guess a project's identity or scope:
- Very specific: line 111
Check every badge's <a href> still points somewhere real.
- Very specific: line 112
Look for prose that should become an admonition, <details>, table, or diagram (references/markdown-patterns.md).
“Common” is measured: kept by at least 6 in 10 of the 5 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.
Files
README craft
Two modes: new (scaffold from scratch) and restyle (upgrade an existing README's presentation). Same conventions either way.
Step 1 — ask badge source
Before writing or touching any badge, ask via AskUserQuestion: shieldcn.dev or shields.io. Never default, never infer from existing badges, never mix sources in one README. Skip only if the user already named the source in this request.
Chosen source can't render a needed badge → tell user and drop it; don't silently fall back to the other source.
Markdown patterns beyond header/badges (admonitions, collapsibles, kbd, mermaid, diff, tables, footnotes, task lists, captions, back-to-top) → references/markdown-patterns.md. Read it before writing README body content.
Header pattern
Centered icon above the title, tagline below:
<p align="center">
<img src="<path-to-icon>" width="64" height="64" alt="">
</p>
<h1 align="center">Project Name</h1>
<p align="center">
One-line description.
</p>Icon: prefer an existing favicon/app-icon asset (PNG/ICO). GitHub strips inline <svg> from rendered READMEs, so even if a .svg logo exists, reference it via <img> (rendered as a raster preview by GitHub) or use a PNG/ICO sibling if one exists. Look in docs/assets/, public/, assets/, or repo root; ask the user for a path if nothing obvious is found — don't guess a project's icon. Separate dark/light icon assets → themed logo pattern in references/markdown-patterns.md.
Badges
Every badge: live or meaningful signal + <a href> to the page it represents (stargazers, npm page, workflow runs, LICENSE, funding page). No bare decorative badges.
shieldcn.dev
Base https://shieldcn.dev. Wrap every badge in <picture> with dark <source> + light <img> fallback:
<a href="<link-target>"><picture><source media="(prefers-color-scheme: dark)" srcset="https://shieldcn.dev/<path>.svg?mode=dark"><img alt="<Label>" src="https://shieldcn.dev/<path>.svg?mode=light"></picture></a>| Badge | Path |
| Stars | github/<owner>/<repo>/stars |
| Release | github/<owner>/<repo>/release |
| License | github/<owner>/<repo>/license |
| Last commit | github/last-commit/<owner>/<repo> |
| Contributors | github/contributors/<owner>/<repo> |
| Release downloads | github/downloads/<owner>/<repo> |
| npm version | npm/<pkg> |
| npm downloads (weekly / monthly) | npm/<pkg>/downloads / npm/dm/<pkg> |
| Homebrew | homebrew/<formula> |
| Discord members | discord/members/<inviteCode> |
| Static | badge/<Label>-<value>-<hex> |
Useful params: variant (default/secondary/outline/ghost/branded), size (xs/sm/default/lg), logo, color. Keep params identical across all badges in a README. Full reference: https://shieldcn.dev/docs/api-reference — check there for any path not in the table; don't guess one.
shields.io
No dark/light wrapping:
<a href="<link-target>"><img src="https://img.shields.io/<path>" alt="<Label>"></a>CI: github/actions/workflow/status/<owner>/<repo>/<file>.yml — targets the real workflow file instead of guessing.
Never add (both sources)
- Vanity — view/visitor counters (incl. shieldcn
views/*), followers, X/Twitter follow, repo size, "PRs welcome", "maintained: yes", "made with love", novelty badges (NBA, Tokscale, flags). - Redundant — info already stated elsewhere in the README or another badge: license badge when a License section exists, version badge when version is in the title/install command, two download badges for the same package, stars badge next to a stars chart.
Static tech-stack badges ("TypeScript", "React") are allowed — one row max, only for the project's core stack.
Metric structurally near-zero for this project (dependents on a leaf app, downloads on an unpublished package) → ask before including.
Verify
Check every metric badge once before shipping — swap .svg for .json (both sources support it) and confirm the value and link are real. shieldcn ci picks a workflow itself → confirm its link points at the intended workflow; wrong one → use a static badge or drop it.
Grouping / ordering
Rows by category, each inside one <p align="center">, rows separated by <br>:
- Social proof — stars, contributors, last commit, CI
- Package registry — version, downloads, license
- Stack — static tech-stack badges
- Funding / links — sponsor (Ko-fi / GitHub Sponsors / Open Collective), homepage, Discord
Empty category → drop the row.
New README checklist (scaffold mode)
Ask before writing if any of these are unclear — don't guess a project's identity or scope:
- Project name, one-line tagline, icon asset path.
- Install method(s) — pick the one true "recommended" path; alternates go in a collapsible.
- Badge set — which badges from the chosen source actually apply after the Never-add filter; whether the project publishes to a package registry (gates that row).
- Core sections: What it does → Install → Usage/Config → How it works → FAQ → Support/Credits/License. Drop sections that don't apply; don't pad to match a template.
Restyle checklist (existing README)
- Convert every existing badge to the chosen source — existing mixed sources don't carry over.
- Drop every badge hit by Never-add; list dropped badges in the final response so the user can object.
- Check every badge's
<a href>still points somewhere real; verify metrics per Verify. - Look for prose that should become an admonition,
<details>, table, or diagram (references/markdown-patterns.md). - Confirm the header follows the icon+heading pattern if the project has a natural icon asset.
Note
This skill file follows compact-markdown style (telegraphic, AI-facing). The README it produces does not — write normal human-facing prose there; this file's terseness doesn't carry over to the output.
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
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.

