TL;DR
Token Compressor is a Model Context Protocol (MCP) server. Compress prompts 40-60% using local LLM + embedding validation. Preserves all conditionals. Install with: uvx token-compressor-mcp. Source: https://github.com/base76-research-lab/token-compressor. It speaks the Model Context Protocol and works with any compatible client (Claude, Cursor, Cline, Windsurf, Warp and more).
Token Compressor
Compress prompts 40-60% using local LLM + embedding validation. Preserves all conditionals.
Install Command
uvx token-compressor-mcpFrequently Asked Questions
Token Compressor is a Model Context Protocol server that Compress prompts 40-60% using local LLM + embedding validation. Preserves all conditionals.
You can install it by running the install command: `uvx token-compressor-mcp`. It then becomes available to any MCP-compatible client such as Claude, Cursor, Cline, Windsurf.
The Token Compressor MCP Server is compatible with any client that speaks the Model Context Protocol — including Claude, Cursor, Cline, Windsurf.
Check the project license on its repository before use.
Reviews
0 reviews· 0.0 average
No reviews yet. Be the first to share your experience.
Leave a review
Rating
Details
- Agent Rank
- 57.6
- License
- Last Updated
- Aug 11, 2026
- Languages
- Python
Client Compatibility
Speaks the Model Context Protocol — works with any compatible client.