How do I use the MCP server with AI coding agents?

The Better i18n MCP (Model Context Protocol) server lets AI coding agents like Claude Code and Cursor manage your translations directly — creating keys, translating content, publishing, and managing content entries.

Installing the MCP server

npx @better-i18n/mcp

Or add it to your MCP configuration file:

{
  "mcpServers": {
    "better-i18n": {
      "command": "npx",
      "args": ["-y", "@better-i18n/mcp"],
      "env": {
        "BETTER_I18N_API_KEY": "sk_your_secret_key"
      }
    }
  }
}

Available tools

Translation key management

Tool	What it does
listProjects	List all your projects
getProject	Get project details (languages, key count)
listKeys	Browse translation keys with filters
createKeys	Create new translation keys with values
updateKeys	Update existing keys or add translations
deleteKeys	Remove keys from your project
publishTranslations	Push translations to the CDN

Content CMS

Tool	What it does
listContentModels	List content model schemas
getContentEntry	Get an entry with all translations
listContentEntries	List entries with filters
createContentEntry	Create a new entry (all languages at once)
updateContentEntry	Update an existing entry
bulkUpdateEntries	Update up to 20 entries in one call
bulkCreateEntries	Create multiple entries at once
publishContentEntry	Publish an entry to the CDN

Example agent prompts

"Add a translation key common.save_button with value 'Save' and translate to Turkish, French, and Spanish"

"Find all untranslated keys in the 'auth' namespace and translate them to German"

"Create a help article about getting started with Better i18n, translate to all project languages, and publish it"

"Check if all keys are in sync with the codebase and fix any missing ones"

Authentication

The MCP server uses your secret API key (sk_...). Keep this secure — don't commit it to git.

For Claude Code, add the key to your .claude/settings.json under env.

For Cursor, add it to the MCP settings under env.

What agents can and cannot do

Agents using the MCP server can:

• Create, read, update, delete translation keys
• Translate content using AI
• Manage content entries and models
• Publish to production CDN

Agents cannot:

• Access billing or payment settings
• Delete projects or organizations
• Invite team members
• Access other users' private data

Logging MCP activity

All MCP operations are logged in your project's Activity log with the label "Agent" so you can see what the AI did.