Open-source platform for 3D AI agents. Turn text or a photo into a rigged, animated GLB avatar, give it an LLM brain, memory and a wallet, and embed it anywhere with one web component. Remote MCP server, x402 payments, Three.js, browser-native.
JavaScript 3d-generationai-agentsanimationavatardigital-humangenerative-aiglbgltfimage-to-3dllmmcpmcp-servermodel-context-protocolsolanatext-to-3dthreejsweb-componentswebglwebxrx402
// GET /api/likeness-bench - the data behind the /likeness-bench internal board.
//
// Sibling of /api/quality-bench, and gated where that one is not. The realism
// bench reads committed JSON files about fixed synthetic prompts, so it is safe
// to serve to anyone. This one reads measurements of avatars built from real
// users' faces, so it goes through the same authorizeOps gate as /api/ops/*:
// a signed-in platform admin, or OPS_SECRET, and in production nothing else.
//
// GET /api/likeness-bench -> { distribution, recent, scorer }
// GET /api/likeness-bench?days=7 -> the same over a 7-day window
// GET /api/likeness-bench?limit=100 -> more rows in the recent table
import { cors, error, json, method, wrap } from './_lib/http.js';
import { authorizeOps } from './_lib/ops-auth.js';
import { likenessDistribution, likenessStoreEnabled, recentLikenessScores } from './_lib/likeness-store.js';
import { SCORER_VERSION, SFACE_SAME_IDENTITY_COSINE, LIKENESS_VIEWS } from './_lib/likeness-score.js';
import { MODELS } from './_lib/face-embed.js';
export default wrap(async (req, res) => {
if (cors(req, res, { methods: 'GET,OPTIONS', credentials: true })) return;
if (!method(req, res, ['GET'])) return;
const auth = await authorizeOps(req);
if (!auth.ok) {
return error(res, 401, 'unauthorized', 'likeness scores are internal: sign in as a platform admin or present x-ops-secret');
}
if (!likenessStoreEnabled()) {
return error(res, 503, 'store_unavailable', 'likeness scores need DATABASE_URL');
}
const url = new URL(req.url, 'http://x');
const days = Number(url.searchParams.get('days')) || 30;
const limit = Number(url.searchParams.get('limit')) || 25;
const [distribution, recent] = await Promise.all([
likenessDistribution({ days }),
recentLikenessScores({ limit }),
]);
// The scorer block travels with the data on purpose: a number on this board
// is meaningless without the instrument that produced it, and the most
// likely future confusion is comparing scores taken by two different model
// versions as if they were the same measurement.
return json(res, 200, {
scorer: {
version: SCORER_VERSION,
detectModel: MODELS.detect.id,
detectLicense: MODELS.detect.license,
embedModel: MODELS.embed.id,
embedLicense: MODELS.embed.license,
sameIdentityCosine: SFACE_SAME_IDENTITY_COSINE,
views: LIKENESS_VIEWS,
},
distribution,
recent,
}, { 'cache-control': 'private, no-store' });
});