Automated README summary
Automatically summarized from the README on 13 Oct 2026; not a run by jevbooks. Run instructions and author-reported outcomes still need verification.
What it is
Vigil is a showcase project for Jev by TypeSafe, a "System One" classifier that returns typed decisions (noul, choice, score) with calibrated probabilities. It benchmarks Jev honestly against rules and a trained XGBoost model for eCommerce fraud detection.
How it uses Jev
Jev receives masked transaction state and answers four typed questions in one request: fraud probability (noul), fraud pattern (choice), risk level (score), and recommended action (choice). A policy turns these answers into allow, step-up, review, or block decisions with reasons.
Primitives:noulchoicescore
Technique worth stealing
Passing fired rules to Jev as evidence lifts account-takeover detection from 60% to 76% and promo abuse from 44% to 75%.
Try it
uv sync && (cd scoring-service && npm install); uv run vigil-sim generate --format jsonl --seed 101 --out data/sim/holdout.jsonl; cd scoring-service; npm run eval:offline -- ../data/sim/holdout.jsonl
Model judgments from README excerpts
Project signals use repository metadata, an existing summary and the first 700 README characters, not the full project. “Not found” means not detected in that material; it does not establish absence. These are the original judgments with recorded models and dates, not source audits or run results.
Models & evaluation dates
- Inclusion gate
jev-1.13.0- Project signals
jev-1.13.0- Category
jev-1.13.0- Pattern tags
jev-1.13.0
Evidence
These signals use repository metadata, an existing summary and at most 700 cleaned README characters. ≥ 0.60 means reported in that excerpt; ≤ 0.40 means not found there. The middle band is inconclusive. None is a source-code audit or a test result. The runnable-tool question excludes research experiments; the recommendation question asks for a learning shortlist, not a quality verdict.
- Jev is core to the projectFound in material0.96
- Shows a System One patternFound in material0.93
- Describes handling uncertaintynot found in the excerpt0.15
- Reports measured numbersunclear0.58
- Is a runnable toolnot found in the excerpt0.18
- Top-10 learning shortlist signalFound in material0.70
- Is a replica of the modelnot found in the excerpt0.08
- Solves a common problem (0–2)score on a 0–2 scale0.99 / 2
- About JevFound in material0.98
no-judgment band (0.40–0.60)
no-judgment band (0.40–0.60)
The Jev model and date for each judgment stage are listed above. Card written by DeepSeek V4.1 Flash from the README on 13 Oct 2026.