Modelwatch runs on your own machine — the client starts it, so there is no endpoint to ping. 59 installs a week from npm. Last commit 26 Oct 2026.
Continuous behavioral drift monitoring for LLM apps — catches silent provider model updates.
We read the source, 6 d ago · rules 995398eba69f
What this server is able to do. For an MCP server this is often the job itself — a terminal server runs commands because that is what it is for. Listed so you know what you are plugging in, not as an accusation.
<script>document.getElementById('post').innerHTML = marked.parse("# Stable week \u2014 no significant drift across the major providers\n\n*May 11, 2026 · ModelWatch Drift Report*\n\nThis week's behavioral check across the major hosted LLMs. Each row is a\ncanary spec \u201…
Is this your server and something here is wrong? Tell us — corrections are free and do not require a plan.
We found places where it runs commands, builds paths or queries from values it is given. None of that is a flaw by itself — it becomes one when the code changes, and code changes quietly between releases. We re-read it on every one.
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 modelwatch -e 'MODELWATCH_API_KEY=<MODELWATCH_API_KEY>' -- npx -y modelwatch-mcp
{
"mcpServers": {
"modelwatch": {
"command": "npx",
"args": [
"-y",
"modelwatch-mcp"
],
"env": {
"MODELWATCH_API_KEY": "<MODELWATCH_API_KEY>"
}
}
}
}
[mcp_servers.modelwatch]
command = "npx"
args = ["-y", "modelwatch-mcp"]
env = { MODELWATCH_API_KEY = "<MODELWATCH_API_KEY>" }
{
"mcpServers": {
"modelwatch": {
"command": "npx",
"args": [
"-y",
"modelwatch-mcp"
],
"env": {
"MODELWATCH_API_KEY": "<MODELWATCH_API_KEY>"
}
}
}
}
{
"servers": {
"modelwatch": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"modelwatch-mcp"
],
"env": {
"MODELWATCH_API_KEY": "<MODELWATCH_API_KEY>"
}
}
}
}
Put your own value in place of
<MODELWATCH_API_KEY> (API key from https://modelwatch.app (free tier: 5 specs, 500 runs/mo)).
The author’s registry entry marks it as required.
Optional settings from the registry entry, left out of the block above:
MODELWATCH_API_BASE (Override the API base URL for self-hosted deployments).
A Model Context Protocol (MCP) server for Prometheus monitoring
Website monitoring for AI assistants: catches pages that look broken even when the site is up.
Uptime monitoring and status pages for apps built with coding agents: status, incidents, setup.
MCP server for AI-driven console application automation and monitoring
Stacks contract monitoring: add a watch in one line, get alerts. Hosted, browser sign-in.
Backlink monitoring with dated fetch evidence; checks and change events continue after the chat.
An MCP server that provides tools for managing and monitoring your Fleet Device Management instance.
Premium agentic endpoint for bias-drift-monitor-mcp.
Answers built from our own checks of this server.