Skip to content
MidLyr Docs

Quick Start

Section titled “Recommended Path”

Start with MCP when you are connecting an AI agent. Use the CLI when you want to test or script the same workflows from a terminal.

MCP

Section titled “MCP”

Connect your MCP client to MidLyr’s remote MCP server:

https://mcp.midlyr.com/mcp

For self-configured clients, set your API key in the environment used by the client:

Terminal window
export MIDLYR_API_KEY="ml_..."

Then add the copy-paste configuration from MCP Configuration and call tools like readRegulatoryContent, startComplianceScreening, and getJob. ChatGPT apps and connectors use OAuth instead of an API-key header.

CLI

Section titled “CLI”

Install the MidLyr CLI:

Terminal window
npm install -g @midlyr/cli

Authenticate (opens a browser to provision an API key):

Terminal window
midlyr login

Submit a screening job. By default the CLI polls until the job succeeds or fails and prints the terminal result:

Terminal window
midlyr screen-analysis \
  --scenario dispute \
  --text "we can't provide a provisional dispute credit until the investigation is completed"

Pass --no-wait to return immediately with the pending job record instead of polling.

SDK

Section titled “SDK”

Install the TypeScript SDK:

Terminal window
npm install @midlyr/sdk-js

Create a client and submit a screening call:

import { Midlyr } from "@midlyr/sdk-js";


const midlyr = new Midlyr({
  apiKey: process.env.MIDLYR_API_KEY!,
});


const response = await midlyr.analysis.screen({
  content: { type: "text", text: "Get 0% APR for life!" },
  scenario: "marketing_asset",
});


const job = await midlyr.jobs.get(response.id);

The SDK uses native fetch, authenticates via the x-api-key header, and has no runtime HTTP dependency.

Next Steps

Section titled “Next Steps”