Skip to content
Metatext Decide — typed decisions for your agents, with a probability on every answerGet early access
Metatext
Sign Up

jev-home-assistant-sentinel

Home Assistant integration that asks Jev, Clef or local Laya about device actions, then applies policy and verifies state separately.

ProjectbetaMIT
View on GitHub ↗

Overview

Jev Home Assistant Sentinel is a Home Assistant integration that adds a safety boundary around AI-assisted decisions about physical devices. It sends a bounded case to a System One decision model, applies a deterministic policy check, and records state verification separately from the command result, so recommendations stay advisory and the evidence is visible.

It can reach several decision models: hosted Jev via TypeSafe or OpenRouter, Cloudflare's Clef and Clef Flash, and local Laya or other compatible engines. Exactly four modes set which side leads and whether the other is a fallback: API with local fallback, API only, local only, and local with API fallback; in the 'only' modes failures are reported, never rerouted. The README stresses that Jev is one member of the System One model category, not its name.

What the decision model decides

TypeQuestionAnswers
Yes / noShould this device action proceed given the case and policy?(inferred)—

How it works

  1. Home Assistant raises a bounded decision case
  2. Selected model (Jev, Clef or Laya) answers per the decision mode
  3. Deterministic policy check gates the action
  4. Verify device state separately from the command result

Strengths

  • Advisory decisions with explicit policy
  • Hosted and local model options with clear fallback rules
  • Separate state verification

Caveats

  • Physical device control carries real-world risk
  • Requires configuring model access

At a glance

Type
Project
Status
beta
Language
Python
Runtime
Home Assistant integration
Has code
yes
License
MIT

Use cases and integrations

How curated lists describe it

  • A safety boundary for AI-assisted Home Assistant decisions, pairing explicit policy checks with chained Jev decision adapters.

    logicrw/awesome-jev-projects

  • A safety boundary for AI-assisted Home Assistant decisions, with explicit policy checks and deterministic state verification.

    hellogumbo/awesome-jev

Frequently asked questions

Which models can it use?

Jev, Cloudflare Clef and Clef Flash, and local Laya or similar engines.

What are the decision modes?

API with local fallback, API only, local only, and local with API fallback.

Does the model control devices directly?

No, a deterministic policy check and separate state verification wrap every decision.

Related projects

  • jev-reasoning-navigator

    PRAXEON runtime supervisor for autonomous agents: actions are checked by policy, a LAYA or TypeSafe semantic judge, signed and sandboxed.

    LibrarybetaPython
  • jev-laya-dsh

    Local Laya decision layer for DeepSeek Harness: routes prompts, gates risky tool calls and swaps models, all offline at no token cost.

    Projectbeta
  • phocinae-server

    Local FastAPI server for the Phocinae 144M decision model implementing the System One /v1/systemone contract on 127.0.0.1.

    ServicebetaPython
  • Strands Decider 2B★ 612

    AWS Strands Labs' 2B decision model: Qwen3.5-2B with a pointer head scoring options and calibrated confidence, plus a CLI.

    ModelbetaPython
  • jev-style

    Open 0.8B and 2B Qwen3.5 decision models with a local systemone-compatible server, MCP tools, agent skills and a Claude Code guard.

    ServicebetaPython
  • Open Spark Jev

    spark-s1: local Jev-inspired decision models on Qwen3.5-4B for NVIDIA DGX Spark, trained on code- and solver-labelled decision rows.

    ModelresearchPython