Agent Gate runs on your own machine — the client starts it, so there is no endpoint to ping. 142 installs a week from npm.
Local mandate gate and proof trail for AI agents that take consequential actions.
Today is the operative word: we check Agent Gate 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-gate -- npx -y forge-agent-gate
{
"mcpServers": {
"agent-gate": {
"args": [
"-y",
"forge-agent-gate"
],
"command": "npx"
}
}
}
[mcp_servers.agent-gate]
command = "npx"
args = ["-y", "forge-agent-gate"]
{
"mcpServers": {
"agent-gate": {
"args": [
"-y",
"forge-agent-gate"
],
"command": "npx"
}
}
}
{
"mcpServers": {
"agent-gate": {
"args": [
"-y",
"forge-agent-gate"
],
"command": "npx"
}
}
}
This one needs environment variables set before it will start:
FORGE_API_KEY (Forge decision-API key (starts with fi_). Set this or FORGE_BEARER_TOKEN.), FORGE_BEARER_TOKEN (Alternative to FORGE_API_KEY: a Forge bearer token for the decision API.), FORGE_API_BASE_URL (Base URL of the Forge decision API. Defaults to the hosted Forge cloud.), FORGE_TENANT_ID (Tenant id stamped into every proof-trail record.), FORGE_RECORD_MODE (required = fail-closed (block if Forge cannot record); best_effort = execute anyway.), AGENT_GATE_MANDATE_PATH (Absolute path to the signed mandate (policy) JSON. Create it with `forge-agent-gate init`.), AGENT_GATE_KILL_FILE (If this file exists, every gated write is hard-blocked (master kill switch).), KALSHI_API_KEY_ID (Kalshi API key id. Only for the live Kalshi venue adapter; stays on your machine.), KALSHI_PRIVATE_KEY_PATH (Path to your Kalshi RSA private key. Only for the Kalshi venue adapter.), KALSHI_ENV (Kalshi environment: demo or prod. Only for the Kalshi venue adapter.).
The author declared them in the registry entry; get the values from the project itself.
Passport, mandate, fail-closed action gate and tamper-evident journal for AI agents.
Local-first project memory and mistake-prevention guardrails for AI coding agents.
Pre-execution policy gate for consequential agent actions with durable trust receipts.
Exact-action approval for consequential agent actions: request, track, and verify signed receipts.
Constitutional guardrails and loop detection for AI agents
Independent effect verification and signed receipts for consequential AI agent actions.
A local receipt and approval gate for AI agent sessions. The agent can act, but it cannot sign.
MUJ428 verifies intent and evidence before consequential AI-agent actions.
Answers built from our own checks of this server.