Token Compressor runs on your own machine — the client starts it, so there is no endpoint to ping. 55 installs a week from pypi. Last commit 8 Mar 2026.
Compress prompts 40-60% using local LLM + embedding validation. Preserves all conditionals.
Nothing serious here today
Today is the operative word: we check Token Compressor every 15 minutes and re-read its code on every release. Watch it and you find out the day that stops being true.
Connect this server
This server runs on your own machine — install it with the package manager and the client starts it for you. Package name taken from the official registry entry.
claude mcp add token-compressor -- uvx token-compressor-mcp{
"mcpServers": {
"token-compressor": {
"args": [
"token-compressor-mcp"
],
"command": "uvx"
}
}
}[mcp_servers.token-compressor]
command = "uvx"
args = ["token-compressor-mcp"]{
"mcpServers": {
"token-compressor": {
"args": [
"token-compressor-mcp"
],
"command": "uvx"
}
}
}{
"mcpServers": {
"token-compressor": {
"args": [
"token-compressor-mcp"
],
"command": "uvx"
}
}
}Alternatives to Token Compressor
same job, measured the same wayRead-only semantic search MCP server for Obsidian vaults using local Ollama embeddings.
Local-first memory for coding agents — MCP server, single SQLite file, local embeddings
Local-first audio analysis: BPM, musical key, production profile, and CLAP vibe embeddings.
Local semantic search over Obsidian vaults via Smart Connections embeddings. Multi-vault, private.
MCP server for Obsidian Smart Connections. Semantic search using your vault's embeddings.
Local FAISS vector database for RAG with document ingestion, semantic search, and MCP prompts.
BM25 search + tree navigation over markdown docs for AI agents. No embeddings, no LLM calls.
Local semantic search — embedding-powered grep for files, zero external services.
Token Compressor — questions
Answers built from our own checks of this server.