Last released Sep 28, 2026
gut's judgment calls as MCP tools: likely, classify, rate and each, answered by small models.
Judgment calls as one line of Python, on small, fast, cheap models. likely / classify / rate, answering YES, NO or UNSURE, on any backend.