MCP I18n Extractor
About
MCP I18n Extractor is an MCP server that extracts i18n (internationalization) strings from JavaScript/TypeScript source files and outputs them as structured JSON files. It is designed for developers working on multilingual applications who need to automate the initial extraction…
Details
- Author
- martincik
- Downloads
- 123
- Categories
- Other
Jump to
- Extracts i18n strings from direct object exports (export default { ... })
- Preserves nested objects and arrays in the output JSON
- Handles template strings with variable interpolation
- Supports strings, numbers, booleans, null, and undefined
- Merges with existing JSON files instead of overwriting
- Replaces source files with a configurable migration message (can be disabled)
Install the npm package (@access-intelligence/mcp-i18n), then configure your MCP client to use the server. The server exposes a single tool operation, extract_i18n, which requires a sourcePath (path to the JS/TS file) and a targetPath (path for the output JSON).
MCP I18n Extractor
A Model Context Protocol (MCP) tool for extracting i18n strings from JavaScript/TypeScript code into JSON files.
Features
- Extracts i18n strings from direct object exports (export default { ... })
- Preserves nested objects and arrays
- Handles template strings with variable interpolation
- Supports various data types (strings, numbers, booleans, null, undefined)
- Merges with existing JSON files when present
- Replaces source files with a migration message (configurable)
Installation
npm install @access-intelligence/mcp-i18n
Usage with MCP
When used via MCP, the tool offers a single operation:
- extract_i18n: Extract i18n strings from source code to a JSON file
Example
extract_i18n(
sourcePath: "/path/to/translations.js",
targetPath: "/path/to/output.json"
)
Environment Variables
- DISABLE_SOURCE_REPLACEMENT: Set to 'true' to prevent replacement of source files after extraction
- WARNING_MESSAGE: Customize the warning message added to replaced source files
Testing
npm test
License
MIT
Sign in to leave a review
Use Google, GitHub, or an email account so ratings stay tied to real people.
No reviews posted yet.
