Agent Guard runs on your own machine — the client starts it, so there is no endpoint to ping. 69 installs a week from npm. Last commit 24 Apr 2026.
Constitutional guardrails and loop detection for AI agents
Today is the operative word: we check Agent 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 agent-guard -- npx -y agent-guard-mcp
{
"mcpServers": {
"agent-guard": {
"command": "npx",
"args": [
"-y",
"agent-guard-mcp"
]
}
}
}
[mcp_servers.agent-guard]
command = "npx"
args = ["-y", "agent-guard-mcp"]
{
"mcpServers": {
"agent-guard": {
"command": "npx",
"args": [
"-y",
"agent-guard-mcp"
]
}
}
}
{
"servers": {
"agent-guard": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"agent-guard-mcp"
]
}
}
}
Constitutional-AI safety guardrails for any LLM — personas, constitutions, adjudication.
Local-first project memory and mistake-prevention guardrails for AI coding agents.
Local-first architecture guardrails and project context for JS/TS AI coding agents.
Security guardrails for AI agent payments
Local-first control plane for coding agents: routing, reservations, and guardrails.
Stateful memory for AI agents. Temporal awareness, error loop detection, and preventive constraints.
Security scanning and threat detection for AI agents
Operational memory for AI agents: code graph, session memory & guardrails. No context rot.
Answers built from our own checks of this server.