Self-hosted, Self-Improving, Federated Everything App — dev management, agent orchestration, persistent mind, digital twin, second brain, and creative pipeline from a single dashboard over Tailscale to port the user into the OS
JavaScript ai-agentsdashboarddeveloper-toolsdigital-twinquantified-selfsecond-brainself-hostedtailscale
/**
* Twin Enrichment Scheduler (Human Activity Tracking Phase 7, #2156).
*
* Backstop daily rollup for the observed taste + chronotype evidence. The
* evidence is normally kept fresh incrementally by each media sync (Spotify /
* YouTube call `refreshTwinEvidenceAfterSync()` after recording events), but a
* machine whose media sources are idle can still accumulate message/calendar
* activity that shifts the chronotype histogram — so we recompute once per local
* day regardless.
*
* ZERO provider calls — aggregation is deterministic and LLM-free, so this runs
* unattended with no user consent (the AI-provider policy only gates the
* explicit "interpret" button, which lives on the Digital Twin UI). The issue's
* design explicitly sanctions running the rollup with no consent "since no
* provider calls," so an initial recompute at boot is fine; thereafter it
* recomputes once per local calendar day.
*/
import { todayInTimezone } from '../lib/timezone.js';
import { getUserTimezone } from './userTimezone.js';
import { aggregateTwinEvidence } from './twinEnrichment.js';
let schedulerInterval = null;
let running = false;
// Local calendar day (YYYY-MM-DD) we last recomputed for. Module-level: a
// restart re-runs at most once (cheap, idempotent) — not worth a state file.
let lastRunDate = null;
const CHECK_INTERVAL_MS = 60 * 60 * 1000; // hourly — a daily rollup needs no finer cadence
async function checkSchedule() {
if (running) return;
const timezone = await getUserTimezone();
const today = todayInTimezone(timezone);
if (today === lastRunDate) return; // already rolled up for this local day
running = true;
try {
await aggregateTwinEvidence();
lastRunDate = today;
} finally {
running = false;
}
}
export function startTwinEnrichmentScheduler() {
if (schedulerInterval) {
console.log('🧭 Twin enrichment scheduler: already running');
return;
}
console.log('🧭 Twin enrichment scheduler: starting');
// Wrapped so an early throw (e.g. DB not ready at boot) can't crash the
// process — this runs outside the request lifecycle.
checkSchedule().catch((err) => console.error(`🧭 Twin enrichment initial rollup failed: ${err.message}`));
schedulerInterval = setInterval(() => {
checkSchedule().catch((err) => console.error(`🧭 Twin enrichment rollup check failed: ${err.message}`));
}, CHECK_INTERVAL_MS);
}