readme-i18n
Use when the user wants to translate a repository README, make a repo multilingual, localize docs, add a language switcher, internationalize the README, or update localized README variants in a GitHub-style repository.
Installation & invocation
1. Crée le fichier sur ta machine :
2. Colle le contenu du SKILL.md ci-dessous, et redémarre Claude Code. Tu peux ensuite l'invoquer manuellement avec :
Claude peut aussi la déclencher automatiquement quand le contexte matche.
Traduit un README de repo en plusieurs langues, ajoute un language switcher, internationalise les variantes localisées.
Contenu de la skill
Localize a repository README.md without breaking the repo mechanics around it.
The default job is translate + wire-up:
- read the source-of-truth README
- create localized sibling files such as
README.zh.md - preserve GitHub-flavored Markdown structure and repo-specific tokens
- add or update a shared language selector near the top of every variant
This skill is for multilingual README workflows, not general website/app i18n.
Inputs
Expect these inputs when available:
- source README path, default
README.md - source language, default inferred or English
- one or more target languages
- optional glossary or do-not-translate list
- optional filename override if the repo already uses a different multilingual naming pattern
If target languages are not named, inspect existing translated files, selectors, filenames, issues, or prior repo conventions. If that still leaves the target languages unclear, ask once. Do not invent target languages.
Defaults And Decision Rules
- Treat the root
README.mdas the source-of-truth unless the user explicitly says otherwise. - If the source language is ambiguous, ask once. Otherwise assume English.
- Keep section order aligned with the source README. Only make a small heading wording adjustment when needed to produce a valid localized anchor.
- Output localized siblings with
README.<bcp47-tag>.mdnaming unless the repo already has a different established pattern that should be preserved. - Update an existing language selector in place. Do not duplicate it.
- Translate only human-language content.
- Preserve project names, package names, commands, CLI flags, option names, environment variables, URLs, file paths, inline code, code fences, HTML attributes, and badge/image URLs.
- Badge alt text or visible labels may be translated only when the change does not require changing the badge URL, query params, or image source.
- If a glossary or do-not-translate list is provided, apply it consistently across every target language.
Workflow
1. Establish the source README and languages
- Confirm the source README path. Default to
README.md. - Identify the source language from the file contents and repo context.
- Determine target languages from the request or existing repo pattern.
- Note any glossary terms, product names, or phrases that must stay untranslated.
2. Audit the Markdown structure before translating
Read the source README once as structure, not prose. Open references/preservation-checklist.md and inventory the elements most likely to break:
- headings and heading levels
- badge rows, shields URLs, and image links
- tables and alignment rows
- raw HTML blocks and inline HTML
- GitHub alerts or admonitions such as
> [!NOTE] - code fences, inline code, commands, and config snippets
- intra-document anchors such as
(#installation) - relative links to files, docs, screenshots, or other README variants
If the README already has localized siblings, inspect them too before choosing filenames or selector style.
3. Translate only the prose layer
Translate:
- paragraph text
- list item prose
- table cell prose
- visible text inside HTML blocks
- image alt text when safe
- selector labels and other human-facing labels
Do not translate:
- fenced code blocks
- inline code spans
- shell commands
- flags such as
--help - env vars such as
OPENAI_API_KEY - URLs
- file paths
- repo/package/project identifiers
- badge and image URLs
When in doubt, preserve the literal token and translate the surrounding sentence instead.
4. Preserve structure while writing the localized README
- Keep the same heading hierarchy and section order as the source.
- Keep the same number of code fences unless the user explicitly asks to rewrite examples.
- Preserve table shape, list nesting, HTML wrappers, and Markdown comments.
- Preserve relative links unless a link intentionally needs to point at a localized sibling README.
5. Rewrite localized anchors and anchor-dependent links
When a translated heading changes, GitHub will generate a different heading ID. After translating headings:
- rewrite every same-file
(#...)link so it matches the localized heading slug in that file - preserve custom explicit anchors such as
<a id="...">unless the file already uses localized explicit IDs - verify that every intra-document anchor target resolves to an existing heading or explicit anchor
Prefer a small heading wording adjustment over a broken anchor. The section order should still match the source README.
6. Write sibling files using the repo's naming pattern
Default to sibling filenames like:
README.zh.mdREADME.es.mdREADME.fr.md
If the repo already uses a different multilingual naming pattern, keep using it consistently rather than forcing the default pattern.
7. Insert or update the language selector
Open references/language-selector-reference.md before editing selectors.
Placement:
- keep the selector near the top of the file
- if the README starts with a title, badges, hero image, or short intro block, place the selector immediately after that opening cluster
Behavior:
- update an existing selector block in place if one already exists
- if you add a new selector, use the canonical marker comments from the reference file so later runs can update it deterministically
- emphasize the current language and link the other variants
- keep selector order and labels consistent across every README variant
8. Final verification
Before finishing, verify:
- localized filenames follow the chosen pattern
- every README variant contains exactly one selector block
- code fence counts are preserved
- badge/image URLs and relative file links still point to the original targets unless intentionally localized
- every
(#...)link resolves inside its own file - the localized README still feels structurally identical to the source
Output
Produce:
- one localized sibling README per target language
- an updated selector block in every README variant
- a brief note to the user covering created or updated files, any assumptions, and any terms intentionally left untranslated
Maintenance Note
Keep README.md as the canonical source unless the user says otherwise. When the source README changes later, update each localized sibling by diffing the changed prose, then re-check selectors, filenames, and anchor links instead of reformatting the whole file from scratch.
Example Prompts
Example 1
Translate this README into Chinese and add a language switcher. Keep badge URLs, code fences, and all commands exactly as they are.
Example 2
Make the repo multilingual. Add Spanish and Chinese README variants, keep the internal anchor links working, and wire the selector into every file.
Example 3
We already have README.zh.md. Add README.es.md and update the existing selector in place instead of adding a second one.
Common Mistakes
- Translating fenced code blocks or inline code instead of only the prose around them
- Duplicating the language selector instead of updating the existing block
- Translating a heading but forgetting to rewrite same-file
(#...)links - Changing badge URLs or image sources while trying to translate visible labels
- Reordering sections in the localized README even though the source README is the authority
Skills proches
better-auth-best-practices
Configure Better Auth server and client, set up database adapters, manage sessions, add plugins, and handle environment variables. Use when users mention Better Auth, betterauth, auth.ts, or need to set up TypeScript authentication with email/password, OAuth, or plugin configuration.
csv-to-fullstack-site
Transforms a CSV, Excel or Google Sheet into a deployed full-stack Next.js site on the user's stack (GitHub + Vercel + Supabase + SendGrid). Use this skill whenever the user has tabular data (any domain - real estate, contacts, recipes, alternance, inventory, anything) and wants to turn it into a live web app, a directory site, an internal tool, or a CRM-style interface. Trigger phrases include "j'ai un CSV", "I have a spreadsheet", "build me a site from this Excel", "fais-moi un site avec ces données", "deploy this Google Sheet as an app", or any time the user uploads tabular data and mentions wanting it online. Encodes the user's specific stack conventions (Next.js 14 App Router, shadcn/ui with @base-ui/react, Tailwind, Zod 4 + react-hook-form, Supabase Postgres + RLS + magic link, Vercel auto-deploy, SendGrid in French) and all the gotchas they hit building production apps. Even if the user only says "build me a site for X" without mentioning the data file explicitly, ask them if they have a CSV/Excel/Sheet and use this skill.
finishing-a-development-branch
Use when implementation is complete, all tests pass, and you need to decide how to integrate the work - guides completion of development work by presenting structured options for merge, PR, or cleanup
firecrawl
| Search, scrape, and interact with the web via the Firecrawl CLI. Use this skill whenever the user wants to search the web, find articles, research a topic, look something up online, scrape a webpage, grab content from a URL, get data from a website, crawl documentation, download a site, or interact with pages that need clicks or logins. Also use when they say "fetch this page", "pull the content from", "get the page at https://", or reference external websites. This provides real-time web search with full page content and interact capabilities — beyond what Claude can do natively with built-in tools. Do NOT trigger for local file operations, git commands, deployments, or code editing tasks.
notion-api-cli
Piloter l'API Notion en bas niveau et via le CLI ntn : créer/mettre à jour des pages, query des databases, déployer des workers Notion, uploader des fichiers, appeler l'API REST Notion. Charge ce skill quand l'utilisateur veut 'appeler l'API Notion', 'déployer un worker Notion', 'uploader un fichier dans Notion', ou toute opération technique bas niveau sur Notion (≠ conception d'espace Notion = skill notion).
shadcn
Manages shadcn components and projects — adding, searching, fixing, debugging, styling, and composing UI. Provides project context, component docs, and usage examples. Applies when working with shadcn/ui, component registries, presets, --preset codes, or any project with a components.json file. Also triggers for "shadcn init", "create an app with --preset", or "switch to --preset".