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
/**
* Oracle — social signal ingestion.
*
* POST /api/oracle/social
* Body: { tweets: XActionsTweet[], network?: string }
*
* Accepts a batch of raw tweets (from XActions, a scraper, or any source),
* extracts coin mentions by $SYMBOL pattern, maps them to known mints in the
* brain, and upserts oracle_narrative with a social-boosted virality score.
* The social virality compounds the existing heuristic virality — it never
* overwrites a higher LLM-derived score.
*
* Tweet shape (XActions format):
* { id, text, createdAt, url, author:{ username }, metrics:{ views, likes, retweets } }
*
* An author in KOL_ACCOUNTS below gets a 1.5× virality multiplier vs an
* anonymous account.
*
* Rate-limited per IP (limits.oracleSocialIp). Auth: optional. Unauthenticated
* calls are accepted so XActions → Oracle works without session overhead.
*/
import { cors, json, method, readJson, wrap, error, rateLimited } from '../_lib/http.js';
import { limits, clientIp } from '../_lib/rate-limit.js';
import { sql, sqlValues, isDbUnavailableError } from '../_lib/db.js';
const NETWORK = 'mainnet';
// $SYMBOL pattern — captures ticker after $, up to 10 chars, word boundary.
const SYMBOL_RE = /\$([A-Z]{2,10})\b/g;
// KOL accounts that get virality multiplier
export const KOL_ACCOUNTS = new Set([
'nichxbt', 'trythreews',
// add more as the platform learns which accounts move prices
]);
function viralityFromTweet(t) {
const views = Number(t.metrics?.views || 0);
const likes = Number(t.metrics?.likes || 0);
const rts = Number(t.metrics?.retweets || 0);
// Log scale: 10k views ≈ 30 pts, 100k ≈ 50, 1M ≈ 70
const viewScore = views > 0 ? Math.min(70, Math.log10(views + 1) * 14) : 0;
// Engagement boost
const engScore = Math.min(20, (likes + rts * 2) / Math.max(1, views) * 1000);
const kol = KOL_ACCOUNTS.has((t.author?.username || '').toLowerCase());
const base = Math.round(viewScore + engScore);
return Math.min(100, kol ? base * 1.5 : base);
}
function extractSymbols(text) {
const out = new Set();
for (const [, sym] of text.toUpperCase().matchAll(SYMBOL_RE)) out.add(sym);
return [...out];
}
export default wrap(async (req, res) => {
if (cors(req, res, { methods: 'POST,OPTIONS', credentials: false })) return;
if (!method(req, res, ['POST'])) return;
const ip = clientIp(req);
const rl = await limits.oracleSocialIp(ip);
if (!rl.success) return rateLimited(res, rl);
const body = await readJson(req);
const tweets = Array.isArray(body?.tweets) ? body.tweets : [];
const network = body?.network === 'devnet' ? 'devnet' : NETWORK;
if (!tweets.length) return error(res, 400, 'no_tweets', 'tweets array required');
if (tweets.length > 500) return error(res, 400, 'too_many', 'max 500 tweets per call');
// Extract all unique symbols across the batch
const symbolMap = new Map(); // symbol → { maxVirality, tweetCount }
for (const t of tweets) {
if (!t?.text) continue;
const virality = viralityFromTweet(t);
for (const sym of extractSymbols(t.text)) {
const cur = symbolMap.get(sym) || { maxVirality: 0, tweetCount: 0, urls: [] };
cur.maxVirality = Math.max(cur.maxVirality, virality);
cur.tweetCount++;
if (t.url && cur.urls.length < 3) cur.urls.push(t.url);
symbolMap.set(sym, cur);
}
}
if (!symbolMap.size) {
return json(res, 200, { ok: true, mints_updated: 0, symbols_found: 0, tweet_count: tweets.length });
}
// Look up mints for the mentioned symbols (case-insensitive match in brain)
const symbols = [...symbolMap.keys()];
const coinRows = await sql`
select mint, symbol from pump_coin_intel
where network = ${network}
and upper(symbol) = any(${symbols})
order by first_seen_at desc
`.catch((err) => {
// An empty lookup answers "no known mints matched the mentioned symbols",
// which tells the caller its batch was useless and should not be retried.
// During a connectivity failure that discards real social signal for good,
// so propagate (wrap() → 503 + Retry-After) and let the sender retry; a
// statement fault still degrades to the documented empty answer.
if (isDbUnavailableError(err)) throw err;
return [];
});
// Dedupe: keep the most recent mint per symbol
const mintBySymbol = new Map();
for (const row of coinRows) {
const sym = row.symbol.toUpperCase();
if (!mintBySymbol.has(sym)) mintBySymbol.set(sym, row.mint);
}
if (!mintBySymbol.size) {
return json(res, 200, { ok: true, mints_updated: 0, symbols_found: symbolMap.size, tweet_count: tweets.length, note: 'no known mints matched the mentioned symbols' });
}
// Upsert oracle_narrative — virality is additive with existing social signal,
// capped at 100. Never downgrades an LLM-sourced row.
const updates = [];
for (const [sym, data] of symbolMap) {
const mint = mintBySymbol.get(sym);
if (!mint) continue;
updates.push({ mint, network, virality: data.maxVirality, tweetCount: data.tweetCount });
}
// Single multi-row upsert instead of one round-trip per symbol — a batch of up
// to 500 tweets can mention dozens of known mints, and N sequential INSERTs
// would serialize the whole request behind DB latency. classified_at is bound
// per row; the ON CONFLICT branch still uses now() for the update path.
// Not guarded: this write IS the request. A swallowed failure used to answer
// `ok: true` with `mints_updated: 0` next to a populated `updated` array: a
// self-contradicting receipt that told the sender its batch had landed when
// nothing was written. wrap() classifies instead (503 on a connectivity
// failure, 500 on a statement fault), so the sender knows to retry.
const rows = updates.map((u) => [u.mint, u.network, u.virality, 0.4, 'heuristic', new Date()]);
await sql`
insert into oracle_narrative (mint, network, virality, confidence, source, classified_at)
values ${sqlValues(rows)}
on conflict (mint, network) do update set
virality = case
when oracle_narrative.source = 'llm' then oracle_narrative.virality
else least(100, greatest(oracle_narrative.virality, excluded.virality))
end,
classified_at = case
when oracle_narrative.source = 'llm' then oracle_narrative.classified_at
else now()
end
`;
return json(res, 200, {
ok: true,
tweet_count: tweets.length,
symbols_found: symbolMap.size,
mints_updated: updates.length,
updated: updates.map(u => ({ mint: u.mint, virality: u.virality, tweets: u.tweetCount })),
});
});