JSON Editor MCP

A Model Context Protocol (MCP) server for editing JSON files with read, write, and deep merge capabilities. Perfect for managing multilingual Next.js projects and other JSON-based configurations.

Note: This is an actively developed tool. Feature requests and bug reports are welcome! File issues on GitHub.

Problems This Solves

• 💰 Significant token savings: Edit specific JSON paths instead of reading/writing entire files
• ⚡ Efficient operations: ~100 tokens per edit vs 4,000+ tokens for full file operations
• 📦 Handles large files: Large translation files may not fit into context windows; targeted operations work regardless of file size
• 🚀 Faster edits: Avoids slow network round-trips from reading/writing entire files
• 🔑 Prevents duplicate keys: AI can't see full translation JSON and creates duplicates; targeted operations avoid this issue
• 🔍 Targeted reads: Read only the values you need using dot notation paths
• ✏️ Targeted writes: Update only specific paths, automatically creates missing nested structures
• 🗑️ Targeted deletes: Remove specific paths from JSON files
• 🔀 Deep merge support: Merge duplicate keys with recursive object merging
• 📚 Multi-file operations: Read, write, or delete the same path from multiple JSON files efficiently
• 📘 TypeScript support: Full type definitions included
• 🤖 MCP compliant: Works with AI assistants (Cursor, Claude, ChatGPT) and development tools

Installation

bun add json-editor-mcp

Usage

MCP Server Configuration

Add to your MCP client configuration:

{
  "mcpServers": {
    "json-editor": {
      "command": "bunx",
      "args": ["json-editor-mcp"]
    }
  }
}

Cursor Rules Integration

Copy the rule file to your project to ensure AI assistants use MCP tools:

cp .cursor/rules/json-editor-mcp.mdc /path/to/your/project/.cursor/rules/

Tools

read_multiple_json_values

Reads the same dot notation path from one or more JSON files in a single operation. Returns a map with file paths as keys and the extracted values as values. Useful for comparing translations across language files or reading from a single file.

Note: For single file operations, pass an array with one file path: ["messages/en.json"]

Input JSON files:

messages/en.json:

{
  "common": {
    "welcome": "Welcome"
  }
}

messages/es.json:

{
  "common": {
    "welcome": "Bienvenido"
  }
}

Tool call:

read_multiple_json_values(["messages/en.json", "messages/es.json"], "common.welcome")

Output:

{
  "messages/en.json": "Welcome",
  "messages/es.json": "Bienvenido"
}

Single file example:

read_multiple_json_values(["messages/en.json"], "common.welcome")

Output:

{
  "messages/en.json": "Welcome"
}

write_json_values

Writes a value to a JSON file at a specified dot notation path. Automatically creates missing nested paths and preserves existing structure.

Input JSON (messages/en.json):

{
  "common": {
    "welcome": "Welcome"
  }
}

Tool call:

write_json_values("/absolute/path/to/messages/en.json", "pages.about.title", "About Us")

Output JSON (messages/en.json):

{
  "common": {
    "welcome": "Welcome"
  },
  "pages": {
    "about": {
      "title": "About Us"
    }
  }
}

Output:

Successfully wrote to /absolute/path/to/messages/en.json

delete_multiple_json_values

Deletes a value at a specified dot notation path from one or more JSON files. Returns a map with file paths as keys and deletion results as values.

Note: For single file operations, pass an array with one file path: ["messages/en.json"]

Input JSON files:

messages/en.json:

{
  "common": {
    "welcome": "Welcome",
    "goodbye": "Goodbye"
  }
}

messages/es.json:

{
  "common": {
    "welcome": "Bienvenido",
    "goodbye": "Adiós"
  }
}

Tool call:

delete_multiple_json_values(["messages/en.json", "messages/es.json"], "common.goodbye")

Output:

{
  "messages/en.json": "Successfully deleted",
  "messages/es.json": "Successfully deleted"
}

Output JSON files:

messages/en.json:

{
  "common": {
    "welcome": "Welcome"
  }
}

messages/es.json:

{
  "common": {
    "welcome": "Bienvenido"
  }
}

merge_duplicate_keys

Performs a deep merge of duplicate keys in a JSON file. Primitives use last-value-wins, objects merge recursively, and arrays use last-value-wins. Useful when AI assistants create duplicate keys because they can't see the full file structure.

Input JSON (messages/en.json):

{
  "common": {
    "welcome": "Welcome"
  },
  "common": {
    "goodbye": "Goodbye"
  }
}

Tool call:

merge_duplicate_keys("messages/en.json")

Output JSON (messages/en.json):

{
  "common": {
    "welcome": "Welcome",
    "goodbye": "Goodbye"
  }
}

API Reference

Path Notation: Dot notation for nested paths (e.g., "common.welcome", "pages.home.title")

Error Handling:

• File not found: Creates empty object {} for reads
• Invalid JSON: Returns error message
• Path not found: Error for reads, auto-creates for writes

Deep Merge: Primitives last-value-wins, objects merge recursively, arrays last-value-wins

Development

git clone https://github.com/peternagy1332/json-editor-mcp.git
cd json-editor-mcp
bun install
bun run build

License

MIT License - see LICENSE file for details.

Support

File issues on GitHub.