MCP Slim Guard runs on your own machine — the client starts it, so there is no endpoint to ping. 34 installs a week from npm. Last commit 14 Aug 2026.
76% fewer MCP tokens in our standard benchmark. Same upstream call. Exact recovery.
Today is the operative word: we check MCP Slim Guard every 15 minutes and re-read its code on every release. Watch it and you find out the day that stops being true.
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 mcp-slim-guard -- npx -y mcp-slim-guard
{
"mcpServers": {
"mcp-slim-guard": {
"args": [
"-y",
"mcp-slim-guard"
],
"command": "npx"
}
}
}
[mcp_servers.mcp-slim-guard]
command = "npx"
args = ["-y", "mcp-slim-guard"]
{
"mcpServers": {
"mcp-slim-guard": {
"args": [
"-y",
"mcp-slim-guard"
],
"command": "npx"
}
}
}
{
"mcpServers": {
"mcp-slim-guard": {
"args": [
"-y",
"mcp-slim-guard"
],
"command": "npx"
}
}
}
Lossless context compression: 2-8x fewer tokens, byte-exact recovery, search inside payloads
Local MCP broker that exposes one compact entry for many upstream servers.
Local MCP broker that exposes one compact entry for many upstream servers.
Benchmarks local LLM inference speed (tokens/sec) on your own hardware via MCP tools.
Exact-decimal double-entry for agents. Call MCP tools; do not do money math in tokens.
Read-only MCP server for the OPERANT AI operating-agent calibration benchmark.
Host your MCP tool over streamable HTTP in one command.
MCP server for OmniStream. Lets AI assistants discover and call any marketplace API.
Answers built from our own checks of this server.