Ratebook
- Role
- Solo build
- Year
- 2026
- Type
- Developer tool
Highlights
- Deterministic, dependency-free rate engine cross-validated against NREL PySAM and a real utility bill (reproduced to within two cents)
- Python and TypeScript engines kept byte-for-byte identical through shared JSON test vectors
- Claude-powered pipeline extracts utility tariff PDFs into a validated, queryable database (6,000+ rates from the public URDB)
- One engine, three surfaces: an MCP server for agents, a Home Assistant integration with live price sensors, and a browser demo
Ratebook turns the messy world of U.S. electricity tariffs — tiered pricing, time-of-use windows, seasonal rates, and fixed charges — into accurate, programmatic bill estimates.
A dependency-free Python rate engine, cross-validated against NREL's PySAM and real utility bills, is mirrored byte-for-byte by a TypeScript port and fed by a tariff database built from the public URDB plus Claude-powered PDF extraction. The same engine ships three ways: an MCP server agents can call, a Home Assistant integration with live price sensors, and a self-contained browser demo.