token-enhancer
by xelektron·★ 64·Score 46
MCP server that strips web pages to clean text, reducing token usage by 86-99%.
Overview
Token Enhancer is a local proxy service that sits between an AI agent and the web, stripping HTML and other noise from pages before they enter the context window. It offers significant token reductions (86-99% as shown in examples like Yahoo Finance) without requiring API keys or external LLM services. The project provides three main modes of operation: as a standalone proxy, as an MCP server with automatic tool discovery, and as a LangChain tool.
Try asking AI
After installing, here are 6 things you can ask your AI assistant:
When to choose this
Choose this when you need to significantly reduce token consumption when processing web content in AI workflows, especially when working with data-heavy sites like financial reports or documentation.
When NOT to choose this
Don't choose this if you need JavaScript-rendered content processing, as it only handles static HTML, JSON, and plain text.
Tools this server exposes
3 tools extracted from the READMEfetch_cleanFetch a URL and return clean text with HTML noise removed.
fetch_clean_batchFetch multiple URLs at once and return clean text for each.
refine_promptClean up a prompt by removing filler words while preserving important information.
Comparable tools
Installation
pip install xelektron-token-enhancerAs an MCP Server (Claude Desktop): Add to your config file:
{
"mcpServers": {
"token-enhancer": {
"command": "python3",
"args": ["-m", "mcp_server"],
"env": {
"REQUESTS_CA_BUNDLE": "/etc/ssl/certs/ca-certificates.crt"
}
}
}
}Remove the env block on macOS/Windows.
FAQ
- Does this service require any external API keys?
- No, Token Enhancer is a local proxy that works without requiring any external API keys or LLM services.
- Which AI clients are compatible with this MCP server?
- The README specifically mentions Claude Desktop, Cursor, and OpenClaw as compatible MCP clients.
- What's the approximate token reduction I can expect?
- According to the documentation, token reduction ranges from 86-99% depending on the source website, with examples showing Yahoo Finance at 99.6%, Wikipedia at 87.4%, Hacker News at 90.1%, and GitHub at 95.9%.
Compare token-enhancer with
Last updated · Auto-generated from public README + GitHub signals.