Skip to main content

i18n with AI

AI coding agents and chat tools have become an essential part of the development workflow, and internationalization is no exception. However, most AI models have limited or outdated knowledge of Lingui's APIs and best practices. This page covers the resources available to give your AI tools accurate, up-to-date Lingui knowledge so they can help you ship localized products faster.

Agent Skills​

Lingui Agent Skills are reusable, procedural knowledge packages for AI coding agents. They provide best practices, common patterns, and troubleshooting guides that agents can follow when working with Lingui - reducing hallucinations and incorrect API usage.

Installation​

The skills follow the Agent Skills format and work with Claude Code, Cursor, Codex, Gemini CLI, GitHub Copilot, and other compatible agents. Use whichever installer matches your tool - each one installs the whole set.

Skills CLI:

npx skills add lingui/skills

Claude Code plugin - all skills load automatically and stay up to date via /plugin marketplace update:

/plugin marketplace add lingui/skills
/plugin install lingui@lingui-skills

Plugins CLI - auto-detects your installed agent tools and installs through each one's native plugin system:

npx plugins add lingui/skills

Gemini CLI extension:

gemini extensions install https://github.com/lingui/skills

GitHub CLI (v2.90+) - installs for GitHub Copilot or any other supported agent. Use --agent <name> to target a specific tool, and gh skill update to pull newer versions:

gh skill install lingui/skills --all

To install a single skill instead of the full set, pass --skill to the Skills CLI:

npx skills add lingui/skills --skill lingui-best-practices

See the Lingui Skills repository for more details about the available skills and their usage.

Context Files​

Lingui follows the llms.txt specification and publishes two files optimized for AI consumption: llms.txt, an index of the documentation pages with a short summary of each, and llms-full.txt, the full docs content in a single streamlined Markdown file. Both are regenerated with every docs build, and some AI tools auto-discover them if you provide https://lingui.dev as a docs source.

While these files are easy to parse, llms-full.txt is large and uses a lot of tokens when loaded directly into context, and a copy saved into your repository goes stale over time. They are best used as a fallback when the AI tool cannot access the latest documentation in other ways. The Context7 MCP server searches the same documentation on demand and returns only the relevant parts, making it the preferred option when available.

tip

Every documentation page is also available as plain Markdown: append .md to its URL, for example https://lingui.dev/ref/macro.md for the Macros reference. A single page costs a fraction of the tokens of llms-full.txt, so when you know which part of the docs the agent needs, give it that URL instead. The links in llms.txt already point to these Markdown versions.

Context7 MCP​

Context7 is an MCP (Model Context Protocol) server that provides AI tools with real-time access to up-to-date library documentation - including Lingui. Unlike static context files, Context7 indexes the latest docs and lets agents search them on demand, keeping token usage low.

Usage​

Add use context7 to any prompt and Context7-enabled agents will automatically pull relevant Lingui documentation:

Add i18n to my React app using Lingui. use context7

MCP Server Configuration​

If your AI tool supports MCP server configuration, you can add Context7 directly:

{
  "mcpServers": {
    "context7": {
      "command": "npx",
      "args": ["-y", "@upstash/context7-mcp"]
    }
  }
}

For tool-specific setup instructions (including remote server and other clients), visit the Context7 page for Lingui.