@mukundakatta/textsanity-mcp
v0.1.1
Published
MCP server: fast unicode/whitespace/encoding cleanup before LLM input. NFKC, zero-width strip, control strip, smart-punctuation conversion, emoji strip. Built for Claude Desktop, Cursor, Cline, Windsurf, and Zed.
Maintainers
Readme
textsanity-mcp
An MCP server that gives AI assistants the ability to clean up messy unicode, whitespace, zero-width characters, and smart punctuation before feeding text to a downstream model.
Works with Claude Desktop, Cursor, Cline, Windsurf, Zed, and any other MCP client.
Operations
- NFKC unicode normalization (
fi→fi, full-width digits → ASCII). - Strip zero-width code points (ZWSP, ZWJ, ZWNJ, BOM, RTL/LTR marks).
- Strip control chars (C0/C1) except
\nand\t. - Collapse whitespace runs to single spaces.
- Trim leading/trailing whitespace.
- ASCII punctuation:
“”‘’–—…→""''-/-.... - Strip emoji (optional).
- ASCII-only mode (optional, applied after smart-punctuation conversion).
Tools exposed
sanitize
Clean a string with the chosen pipeline.
{
"text": " Hello World — with fi and smart “quotes” ",
"ascii_punctuation": true
}→
{ "clean": "Hello World - with fi and smart \"quotes\"" }Configure your MCP client
Claude Desktop's claude_desktop_config.json:
{
"mcpServers": {
"textsanity": {
"command": "npx",
"args": ["-y", "@mukundakatta/textsanity-mcp"]
}
}
}License
MIT.
