ApprovalGate

Human-in-the-loop for AI agents. Pause any agent action, collect a human approval (or rejection), and resume — all through a simple SDK call.

Quickstart

Install the SDK, then wrap risky actions in five lines of code:

bash

# Install
pip install luxkern          # Python
npm install @luxkern/sdk     # Node.js

python

"text-[var(--color-info)] font-medium">from luxkern "text-[var(--color-info)] font-medium">import ApprovalGate
 
gate = ApprovalGate(api_key="lxk_live_xxx")
decision = gate.request(
    action="delete_user",
    context={"user_id": "usr_42", "reason": "Agent flagged ">as spam"},
)
 
"text-[var(--color-info)] font-medium">if decision.approved:
    delete_user("usr_42")

"text-[var(--color-info)] font-medium">import { ApprovalGate } "text-[var(--color-info)] font-medium">from "@luxkern/sdk";
 
"text-[var(--color-info)] font-medium">const gate = "text-[var(--color-info)] font-medium">new ApprovalGate({ apiKey: "lxk_live_xxx" });
"text-[var(--color-info)] font-medium">const decision = "text-[var(--color-info)] font-medium">await gate.request({
  action: "delete_user",
  context: { userId: "usr_42", reason: "Agent flagged as spam" },
});
 
"text-[var(--color-info)] font-medium">if (decision.approved) {
  "text-[var(--color-info)] font-medium">await deleteUser("usr_42");
}

What you get

Slack, email, and dashboard approval channels
Configurable timeout and auto-deny policies
Full audit trail of every approval decision
Bulk approval for batch agent operations

Open ApprovalGate Dashboard · Full API Reference