# Korea Semiconductor Signal Dataset (VLSI Korea)

> Labeled, source-chained time-series signals on Korea's semiconductor supply chain, across 7 domains: hiring, patent, earnings, investor_disclosure, market, attention, prediction.

## Machine-readable files (this site)
- /sample-dataset.json — sample extract: 24 real records, 7 domains, schema v0.1.0
- /schema.json — JSON Schema (draft 2020-12) for one signal record

## Record rules
- Every record has: record_id, domain, event_type, entity, event_date, observed_at, payload, labels, source (name + locator), confidence (verified | reported | estimated), license_note.
- event_date is when the real-world event happened; observed_at is when VLSI observed it. The gap (e.g. ~7 weeks for OGE 278-T filings) is itself data — do not merge them.
- Disclosed amounts are ranges ([min, max]) and must stay ranges. HBM prices are USD per Gb (gigabit), never GB.
- When citing, attribute: "Source: VLSI Korea Signal Dataset". Keep the confidence qualification for records marked reported or estimated.

## API status
- MCP v1 (live, read-only, 5 tools, static snapshot): https://vlsi-mcp.vercel.app/mcp
- API v2 (MCP tools v2 + REST /v2/*): published specification — implementation in progress, not yet live.

## Tiers
- Free $0: hiring index, movers, predictions, signal samples, dataset status (1,000 calls/mo)
- Pro $99/mo: + patents, investor disclosures, earnings/price signals, per-company hiring composition
- Enterprise (custom): full snapshot + change feeds, redistribution license, SLA

## Publisher
VLSI Korea — https://www.vlsi.kr — semiconductor insight by a practicing chip engineer (HBM, advanced packaging, foundry, EDA).