Model Context Protocol

AI-Native Translation Workflow

Give your AI coding agent (Cursor, Claude Desktop, Copilot, or Kiro) direct access to sync translations, search keys, and manage your localization workflow without leaving the editor.

CursorClaude DesktopGitHub CopilotKiroClineWindsurf

What is MCP?

The Model Context Protocol (MCP) is an open standard that lets AI assistants interact with external tools and services. Instead of copy-pasting commands, your AI agent calls Localingos tools directly: syncing translations, searching keys, and pushing updates in the flow of your conversation.

Available Tools

Core tools (local + remote)
Tool
Description
localingos_sync
Push source strings and pull back translations. Only changed content is transferred — uses content hashing to skip unchanged items and only returns translations the client needs.
localingos_push
Push source strings to Localingos without pulling translations.
localingos_pull
Pull latest translations and write locale files to your project.
localingos_status
Show project status: project, locales, source key count, translation files.
localingos_search
Search translation keys or text across your source strings.
localingos_extract
Run the i18n extraction script to rebuild JSON from .messages.ts files.
Additional tools (remote + CLI mcp-serve)
Tool
Description
localingos_get_translatables
Get all source strings for a project.
localingos_get_translations
Get all translations for a project.
localingos_get_translation
Get translations for a specific key.
localingos_translation_jobs
List translation jobs for a project.
localingos_retry_job
Retry a single failed translation job.
localingos_retry_all_jobs
Retry all failed and blocked jobs.
localingos_project_members
List members of a project.
localingos_add_member
Add a member to a project.
localingos_remove_member
Remove a member from a project.
localingos_billing_status
Show billing status, plan, and usage.
localingos_attach_billing
Attach your billing account to a project.
localingos_detach_billing
Detach billing from a project.

Local Setup (stdio)

Install the MCP server and configure your AI agent to spawn it locally. Reads your project config. No server costs.
Install the MCP server
npm install -g localingos
Add to ~/Library/Application Support/Claude/claude_desktop_config.json:
{
  "mcpServers": {
    "localingos": {
      "command": "localingos", "args": ["mcp-serve"],
      "cwd": "/path/to/your/project"
    }
  }
}
Add to .cursor/mcp.json in your project:
{
  "mcpServers": {
    "localingos": {
      "command": "localingos", "args": ["mcp-serve"]
    }
  }
}
Add to .kiro/settings/mcp.json:
{
  "mcpServers": {
    "localingos": {
      "command": "localingos", "args": ["mcp-serve"]
    }
  }
}
In VS Code, open Cline → MCP Servers → Add Local Server:
Command: localingos
Arguments: mcp-serve

Remote Setup (mcp.localingos.com)

No installation needed. Point your agent to our hosted MCP server. Pass your API key and project ID with each tool call.
{
  "mcpServers": {
    "localingos": {
      "type": "streamable-http",
      "url": "https://mcp.localingos.com"
    }
  }
}
{
  "mcpServers": {
    "localingos": {
      "type": "streamable-http",
      "url": "https://mcp.localingos.com"
    }
  }
}
Add to .kiro/settings/mcp.json:
{
  "mcpServers": {
    "localingos": {
      "type": "streamable-http",
      "url": "https://mcp.localingos.com"
    }
  }
}
In VS Code, open Cline → MCP Servers → Add Remote Server:
Server URL: https://mcp.localingos.com
Server Name: localingos
When using the remote server, your AI agent passes your API key and family ID with each tool call — no local config files needed.

Example Conversations

"Sync my translations"
→ Agent calls localingos_sync → pushes 770 strings, pulls 8 locales, writes files
"Are there translations for the billing page?"
→ Agent calls localingos_search with "billing" → returns matching keys and text
"Run the i18n extraction and check for orphans"
→ Agent calls localingos_extract with check=true → reports key counts and any orphaned messages