A quant engine reads the limit order book second by second and turns order-flow imbalance, micro-price, a fair-value model and the shape of the forming candle into a plain-language market state.
That state goes to Jev as 12 atomic questions in one parallel call, including the trade decision itself asked three different ways just to check it agrees with itself, and the bot only fires when the answers are confident and agree with each other and with the flow.
A `jev-bot demo` command writes a self-contained HTML replay showing price against fair value, the order book, every typed answer Jev gave, and the exact state it was sent, so a run can be reviewed after the fact rather than trusted blind.
What you get
- Full Python source code under the MIT license
- The quant feature engine, the 12-question Jev gate, and the HTML replay dashboard






