Building a gRPC Guardian + Intel API on a Prediction OS

What shipped in v0.57

We just cut v0.57 of Kairon Forge — the B2B AI agent platform that ships every agent pre-loaded with prediction-market intelligence.

Guardian gRPC server

The Guardian audit layer is now a dedicated gRPC server. SecurityScanner kernel runs inside, applying rules against each incoming audit record. Five unit tests cover the critical paths.

Intel API real impl

macro_snapshot runs on a 4-hour cron against live Polymarket data. anomaly_detect uses z-score over a configurable rolling window. forecast_calibrated combines market probabilities with historical calibration curves for confidence-banded predictions.

@kairon/sdk v0.0.1

import { KaironClient } from "@kairon/sdk";
const client = new KaironClient({ apiKey: process.env.KAIRON_API_KEY, tier: "pro" });
const snapshot = await client.intel.macroSnapshot({ date: "2026-05-18" });

MCP server (@modelcontextprotocol/server-kairon v0.0.1) exposes the same Intel tools.

Try it

npm install @kairon/sdk @modelcontextprotocol/server-kairon

Forge: kairon.trade/forge