What it is
A demo personal finance tracker with a Python/FastAPI backend and React/TypeScript frontend. It categorizes transactions, detects subscriptions and duplicates, and flags anomalies using Jev's calibrated probabilities, then applies deterministic budget rules.
How it uses Jev
On each transaction, Jev runs: choice to pick a category with confidence, noul for subscription and duplicate probability, and score for 1-5 anomaly rating. Results gate application logic: categories below 0.80 confidence need review, subscriptions at ≥0.70 register, duplicates at ≥0.80 flag.
Primitives:choicenoulscore
Technique worth stealing
Threshold-gating Jev's calibrated probabilities for human review instead of trusting the model's top choice.
Try it
Set JEV_API_KEY in .env, then run 'make setup' and 'make start' from the repository root.
Evidence
Each line is one question put to Jev about the README. ≥ 0.60 reads as yes, ≤ 0.40 as no; in between Jev is not making a call.
- Jev-centricyes0.81
- Shows a System One patternyes0.79
- Handles uncertaintyno0.08
- Measuredno0.06
- Runnableyes0.81
- Worth recommendingunclear0.51
- Model replicano0.03
- Problem scopescore on a 0–2 scale0.21
- About Jevyes0.98
Signals by Jev jev-1.13.0, card written by DeepSeek V4.1 Flash from the README on 20 Sept 2026.