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
// Output shaping for alert rules and fired alerts. The three.ws API already
// returns clean, documented objects; these helpers pin the agent-facing shape so
// it stays stable regardless of incidental fields the API adds, and add a
// one-line human summary to each fired alert.
/**
* Normalize a rule row from /api/alerts/rules into the agent-facing shape.
* Field names mirror the API and the create/update tool inputs one-for-one.
* @param {Record<string, any> | null | undefined} r
*/
export function shapeRule(r) {
if (!r || typeof r !== 'object') return null;
return {
id: r.id,
kind: r.kind,
// `label` is what the user typed (null when auto-named); `label_display`
// is the resolved name the dashboard shows (a derived default when unnamed).
label: r.label ?? null,
label_display: r.label_display ?? null,
target_mint: r.target_mint ?? null,
target_agent: r.target_agent ?? null,
threshold: r.threshold ?? null,
filters: r.filters ?? null,
deliver_in_app: r.deliver_in_app ?? null,
webhook_url: r.webhook_url ?? null,
// Per-rule HMAC secret for verifying the `webhook-signature` header on
// inbound deliveries. Returned only to the rule's owner.
webhook_secret: r.webhook_secret ?? null,
telegram_chat: r.telegram_chat ?? null,
cooldown_seconds: r.cooldown_seconds ?? null,
enabled: r.enabled ?? null,
last_fired_at: r.last_fired_at ?? null,
recent_failures: r.recent_failures ?? 0,
recent_deliveries: Array.isArray(r.recent_deliveries) ? r.recent_deliveries : [],
created_at: r.created_at ?? null,
updated_at: r.updated_at ?? null,
};
}
/**
* Normalize a `pump_alert` notification row from /api/notifications into a
* fired-alert entry. The notification's `payload` is the alert itself (kind,
* mint, amounts, market cap, tx, …); we surface those fields and add a
* human-readable summary.
* @param {{ id?: any, payload?: Record<string, any>, read_at?: any, created_at?: any }} n
*/
export function shapeAlert(n) {
if (!n || typeof n !== 'object') return null;
const payload = n.payload && typeof n.payload === 'object' ? n.payload : {};
return {
id: n.id,
fired_at: n.created_at ?? payload.at ?? null,
read: n.read_at != null,
summary: summarizeAlert(payload),
rule_id: payload.rule_id ?? null,
kind: payload.kind ?? null,
mint: payload.mint ?? null,
name: payload.name ?? null,
symbol: payload.symbol ?? null,
event: payload,
};
}
const usd = (v) =>
v != null && Number.isFinite(Number(v)) ? `$${Math.round(Number(v)).toLocaleString('en-US')}` : null;
/**
* One-line, agent-readable summary of a fired-alert payload. Mirrors the site's
* own alert wording so history reads the same as the in-app feed.
* @param {Record<string, any>} p
*/
export function summarizeAlert(p) {
const tok = p.symbol
? `$${p.symbol}`
: p.name || (p.mint ? `${String(p.mint).slice(0, 4)}…${String(p.mint).slice(-4)}` : 'token');
switch (p.kind) {
case 'graduation': {
const mc = usd(p.market_cap_usd);
return `${tok} graduated to AMM${mc ? ` at ${mc} mcap` : ''}`;
}
case 'new_mint':
return `${tok} just launched`;
case 'whale_buy': {
const sol = p.amount_sol != null ? `${Number(p.amount_sol).toFixed(2)} SOL` : 'a large buy';
const u = usd(p.amount_usd);
return `Whale bought ${sol}${u ? ` (${u})` : ''} of ${tok}`;
}
case 'price_above':
return `${tok} mcap rose above ${usd(p.threshold_usd)} (now ${usd(p.market_cap_usd)})`;
case 'price_below':
return `${tok} mcap fell below ${usd(p.threshold_usd)} (now ${usd(p.market_cap_usd)})`;
case 'market_anomaly':
return p.summary || `${p.subject || 'market'} ${p.processor || ''} anomaly`.replace(/\s+/g, ' ').trim();
default:
return `${tok} alert`;
}
}