@rog0x/mcp-regex-tools

mcp-regex-tools

Regex and text processing tools for AI agents via the Model Context Protocol (MCP).

Tools

Tool	Description
regex_test	Test a regex pattern against text. Returns all matches with indices, captured groups, and named groups.
regex_replace	Find and replace using regex with support for capture group references ($1, $2, $&, $<name>).
regex_extract	Extract all matches of a pattern from text as a structured array. Optionally return only a specific capture group.
regex_explain	Parse a regex pattern and return a human-readable explanation of each component and flag.
text_transform	Apply common text transformations: camelCase, snake_case, kebab-case, PascalCase, UPPER, lower, title case, reverse, trim, deduplicate lines, sort lines, remove blank lines, count words/chars/lines. Chain multiple transforms in one call.

Setup

npm install
npm run build

Usage with Claude Desktop

Add this to your Claude Desktop config file:

• Windows: %APPDATA%\Claude\claude_desktop_config.json
• macOS: ~/Library/Application Support/Claude/claude_desktop_config.json

{
  "mcpServers": {
    "regex-tools": {
      "command": "node",
      "args": ["D:/products/mcp-servers/mcp-regex-tools/dist/index.js"]
    }
  }
}

Usage with Claude Code

claude mcp add regex-tools node D:/products/mcp-servers/mcp-regex-tools/dist/index.js

Or add to your .mcp.json:

{
  "mcpServers": {
    "regex-tools": {
      "command": "node",
      "args": ["D:/products/mcp-servers/mcp-regex-tools/dist/index.js"]
    }
  }
}

Examples

Test a regex

regex_test({ pattern: "(\\d{4})-(\\d{2})-(\\d{2})", text: "Today is 2026-03-19 and tomorrow is 2026-03-20" })

Replace with capture groups

regex_replace({ pattern: "(\\w+)\\s(\\w+)", replacement: "$2 $1", text: "hello world" })

Extract matches

regex_extract({ pattern: "[\\w.]+@[\\w.]+", text: "Contact us at [e-mail] or [e-mail]" })

Explain a regex

regex_explain({ pattern: "^(?:https?://)?(?:www\\.)?([\\w-]+)\\.\\w{2,}$", flags: "i" })

Transform text

text_transform({ text: "hello world example", transforms: ["PascalCase"] })
text_transform({ text: "some text\nwith\nlines\nwith\nduplicates\nlines", transforms: ["deduplicate_lines", "sort_lines"] })

License

MIT