GenderAPI.io plugin

• A key is optional, so Corsair does not prompt for one. With no key connected, GenderAPI.io falls back to an IP-scoped trial of 10 credits per 24 hours, shared by every caller behind the same public IP (on a shared server, every tenant without a key). Connect the key with a connect link or the key option before paid use. meta.access.mode says which one was used: api_key or ip_trial. A missing or unknown key falls back too, while a disabled, expired or restricted key does not and answers 403. A trial batch takes at most 10 items.
• An unknown gender is a billed result, not an error. The answer is HTTP 200 with gender: null, result_status: "unknown" and a reason such as not_found or ambiguous.
• source says who answered. dataset means the name was in the database, ai means a model inferred it, and none means nothing could. confidence_kind follows: observed_frequency for the dataset, model_reported for AI.
• What a call costs. A single request defaults to ai_mode: 'fallback', which costs 1 credit in total whether the dataset or AI answers. ai_mode: 'always' costs 2, and so does a forceToGenderize call that AI answers. A batch defaults to off, so pass ai_mode (for the batch or per item) when a batch needs AI. Phone validation costs 1, and so does an unknown result.
• A batch item can fail without failing the batch. Only successful items are billed, and meta.summary counts total, succeeded, unknown and failed.
• A failed prediction may already be charged. The error catalog says to inspect billing_status before retrying, because a retry starts a new billed operation. The plugin does not retry a prediction or phone call after a 5xx, a timeout or a dropped connection; an HTTP failure is core’s ApiError, whose body keeps meta.usage.billing_status when the API sent one (genderApiIoErrorInfo(error).billingStatus reads it). Only the free GET routes and a 429 are retried.
• usage, capabilities and errorCatalog are free. They are safe to poll: the first gives the balance, the second the batch limit and AI availability, the third every error code with what to do about it.
• Confidence is not a calibrated probability. The docs say so explicitly, so treat it as a ranking signal rather than a certainty.

• 6 typed API operations

Setup

npm install corsair @corsair-dev/genderapiio better-sqlite3
npm install --save-dev @types/better-sqlite3

yarn add corsair @corsair-dev/genderapiio better-sqlite3
yarn add --dev @types/better-sqlite3

pnpm install corsair @corsair-dev/genderapiio better-sqlite3
pnpm install --save-dev @types/better-sqlite3

bun add corsair @corsair-dev/genderapiio better-sqlite3
bun add --dev @types/better-sqlite3

import Database from 'better-sqlite3';
import { createCorsair } from 'corsair';
import { genderapiio } from '@corsair-dev/genderapiio';

export const corsair = createCorsair({
	plugins: [
		genderapiio(),
	],
	database: new Database('corsair.db'),
	kek: process.env.CORSAIR_KEK!,
	multiTenancy: true,
	hub: {
		projectApiKey: process.env.CORSAIR_API_KEY!,
		signingSecret: process.env.CORSAIR_SIGNING_SECRET!,
	},
});

genderapiio()

const { connectUrl } = await corsair.manage.connect.createLink({
	plugin: 'genderapiio',
	tenantId: 'acme',
});
// redirect the user's browser to connectUrl

Example API calls

const tenant = corsair.withTenant('acme');
await tenant.genderapiio.api.phone.validate({ number: '+14155552671', country: 'US' });

const tenant = corsair.withTenant('acme');
await tenant.genderapiio.api.gender.batch({
	items: [{ type: 'name', value: 'Sophia' }, { type: 'name', value: 'Michael', country: 'US' }],
});

What’s next