v0.3.0 — Railway Research Runtime

What changed

• Runs the Python research API alongside the React dashboard in the Railway container.
• Adds authenticated MCX CSV import, clean dataset merge, research execution, and context refresh endpoints.
• Adds a Railway operator CLI, persistent volume guidance, and API integration coverage.
• Keeps demo data explicitly labeled until official MCX observations are imported.

Verification

• 39 Python tests passed.
• React production build and lint passed.
• 5 frontend tests passed.
• Railway deployment itself remains unverified because no Railway project is linked in the build environment; attach a volume at /app/data and set AURUM_ADMIN_TOKEN.

See deployment guide.

Aurum Spread v0.1.0 — Research Prototype

First Hack in Hills ’26 release

Aurum Spread is a cost-aware relative-value intelligence prototype for the Commodity Derivatives Intelligence track.

Highlights

• Interactive monitor for GOLDM, GOLDTEN, GOLDGUINEA, and GOLDPETAL
• Quotation-base and purity normalization
• Expiry, liquidity, statistical, and cost gates
• Lagged walk-forward analytics with five automated tests
• Explicit separation of gross convergence, costs, and gold exposure
• Railway-ready Docker/Caddy deployment with /health
• Security policy, MIT license, contribution templates, and research documentation

Verification

• Production TypeScript/Vite build: passed
• Oxlint: passed
• Python analytics tests: 5/5 passed
• npm high-severity audit: 0 vulnerabilities

Research status

The visible observations and performance metrics are synthetic demo fixtures. This release does not yet claim an empirical MCX edge and does not execute orders or provide investment advice. The next milestone is official Bhavcopy ingestion and a frozen chronological holdout.