Hockeyclaw Kickoff: Agentic NHL Data MVP

Mar 4, 2026

Quick project update: we just kicked off Hockeyclaw, a focused MVP for agentic NHL data workflows.

Project status is tracked on the Projects page .

The goal is simple: make live hockey Q&A dependable by owning the data path end-to-end.

What we’re building

Hockeyclaw is an ingestion + query system that:

That means less guessing, fewer stale answers, and faster responses when someone asks things like:

MVP stack (speed-first)

We’re shipping MVP in Node + TypeScript + Postgres so we can move quickly and validate usage patterns.

For scale, the plan is to keep interfaces clean so a future Go ingestion plane can slot in without a rewrite of the query surface.

Shipped already

Next up is standings movement and nightly recap automation.

Why this matters

The important shift is operational: no more one-off scraping scripts or ad-hoc lookups during chat.

Hockeyclaw turns “can you check that game real quick?” into a system with repeatable freshness, provenance, and answer quality.

I’ll keep posting receipts as each batch lands.