jevbooks

← All projects

jev-router

hyspacex/jev-router

Route OpenAI-style chat requests to a model and reasoning effort via Jev.

Classification & Routing1.00Runner-up: Verification & GuardrailsFail open, fail closed
Automated README summary

Automatically summarized from the README on 28 Oct 2026; not a run by jevbooks. Run instructions and author-reported outcomes still need verification.

What it is

A self-hosted model router for coding agents and OpenAI-compatible clients. It assesses each new task with TypeSafe's Jev, then applies YAML rules to pick a model and reasoning effort from your upstream's pool, binding a session to one model.

How it uses Jev

On a fresh session resolve, Jev receives a bounded state of request excerpts and computed facts and assesses the task. Its result feeds YAML policy to select model, provider, effort, and limits, returned as a binding. If Jev fails, strict admission uses the alias's admission_fallback or refuses with NO_SAFE_ADMISSION.

Primitives:choicescorenoul

Technique worth stealing

Resolve the model once per session before the client sizes its history, then keep the binding stable across the tool loop.

Try it

uv sync --locked, edit quickstart.local.yaml, then uv run jev-router -c quickstart.local.yaml serve.

View on GitHub

Start using Jev: API key and your first request →

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-1.14.0
Project signals
jev-1.14.0
Category
jev-1.14.0
Pattern tags
jev-1.14.0

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 material0.94
  • Shows a System One patternnot found in the excerpt0.21
  • Describes handling uncertaintynot found in the excerpt0.04
  • Reports measured numbersnot found in the excerpt0.02
  • Is a runnable toolFound in material0.97
  • Top-10 learning shortlist signalnot found in the excerpt0.33
  • Is a replica of the modelnot found in the excerpt0.04
  • Solves a common problem (0–2)score on a 0–2 scale1.60 / 2
  • About JevFound in material0.98
Signals
Jev is core to the project0.94Shows a System One pattern0.21Describes handling uncertainty0.04Reports measured numbers0.02Is a runnable tool0.97Top-10 learning shortlist signal0.33Is a replica of the model0.0400.51Solves a common problem (0–2)1.60 / 2012

no-judgment band (0.40–0.60)

Design patterns
Fail open, fail closed0.96Ghostwriter0.58Flight recorder0.56Viewfinder0.53Funnel0.52Thermostat0.4500.51

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 28 Oct 2026.