typesafe-jev-incident-router
Data-incident router and use-case guide: registry facts first, then Jev Choice, Score and Noul with confidence-gated human review.
Overview
This demonstration repository's main value is a guide to Jev use cases for data teams across data engineering, analytics engineering, analytics and data science. A runnable Python example routes synthetic data incidents while keeping exact facts, model judgments, application policy and human review separate.
A registry of known assets is checked first: if the incident names an asset with an authoritative owner, that fact is used and Jev is not asked to infer ownership. Otherwise Jev suggests a response team (Choice), places business impact on an ordered rubric (Score) and estimates whether critical work is blocked (Noul). Named policy constants decide what happens next: Choice confidence below 0.75 or Score confidence below 0.70 goes to review, and a blocked-work probability of 0.70 or higher triggers escalation. The README also offers a simple heuristic for when to use code, Jev, a generative model or a human.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Choice | Which response team should handle this incident? | — |
| Score | What is the business impact of this incident? | — |
| Yes / no | Does the incident report say critical work is blocked? | yes · no |
How it works
- Look up the asset in the registry; use the owner if known.
- Otherwise ask Jev three questions about the incident.
- Apply policy thresholds (0.75 and 0.70) to route or send to review.
- Hand typed decisions and evidence to a human-led workflow.
Strengths
- Exact facts are never delegated to the model.
- Clear, named policy constants.
- Useful guide for data teams.
Caveats
- Synthetic incidents only.
- Demonstration rather than a deployable service.
At a glance
- Type
- Project
- Status
- prototype
- Language
- Python
- Runtime
- scripts
- Has code
- yes
Use cases and integrations
How curated lists describe it
Confidence-gated incident routing with TypeSafe Jev
logicrw/awesome-jev-projects
Frequently asked questions
When is Jev not asked?
When the registry already names an authoritative owner for the asset.
What sends an incident to review?
Choice confidence below 0.75 or Score confidence below 0.70.
What is the main deliverable?
The use-case guide for data teams, with the router as an example.
Related projects
Two small Python apps: Jev triages failed pipeline tasks into review, retry, page or ticket, and routes support messages to agents.
ProjectprototypePythonTwo beginner Python apps using Jev: a message sorter and a support ticket router that send uncertain items to human review.
ProjectprototypePythonPython sandbox turning Jev issue-category and reproduction judgments into testable routing proposals, with offline replay.
ProjectprototypePythonR&D repo with three Jev pipelines (ticket triage, alert triage, deploy-risk gate) built from four reusable patterns.
ProjectresearchPythonFifteen runnable Node recipes using Jev via OpenRouter for triage, dedupe, PII scanning, reranking and a browser agent.
ArticlebetaJavaScriptClaude Code skill that hands bulk 'read a lot to decide a little' judgments over files, logs and tickets to Jev and returns a shortlist.
Agent skillbetaJavaScript