Skip to content

Jev Layer

MCP layer that lets an agent harness ask Jev for a bounded choice from a candidate set it already allows, while permissions and execution stay with the host.

Stars
6
Language
JavaScript
License
MIT
Published by
Community
Author
typakon4
Last verified
Oct 2026

About this project

Let an agent make a bounded choice without handing it control. Jev recommends. Your harness still checks permissions and executes. Hermes · OMP · Codex · generic MCP. Install: npm install --global jev-layer · Integrate a harness · Security model. Jev never executes a selected capability. If it is disabled, unavailable, invalid, or inconclusive, control returns to the host's normal path. Requires Node.js 20 or newer. There are no mandatory runtime dependencies.

Quoted from the project's README. Read the full README.

Highlights

  • Jev never executes a chosen capability; disabled, unavailable, invalid or inconclusive means control falls back to the host's own path.
  • The default demo provider is deterministic and works offline, so routing and the stdio MCP server run with no key and no dependencies.
  • Optional surfaces are off by default and enabled per flag, covering supervision, shadow compaction, context filtering and a browser fast-path.
  • Outcomes can be recorded against the original correlation_id and replayed offline, so routing cases stay reviewable after the fact.

Watch out

The project states plainly that it is not a security boundary: host permissions and approvals remain authoritative and Jev never widens them. Schema, MCP tool, receipt, replay and adapter contracts are all version 1, so changes are meant to be additive and silence there is a break. The browser fast-path is experimental, validated against current real-browser fixtures, with no speedup claim made. Provider-backed tests are excluded from ordinary CI, so only the deterministic demo path runs on every push. Copy reviewed against the repository README on 2026-10-04, not from running it.

Questions about this entry

Short answers drawn from the linked project or page.

More Jev resources like this

Task plugin that puts TypeSafe's System One endpoint behind a QuantumNous/new-api channel, so clients keep posting `/v1/systemone` while new-api handles auth, routing, model mapping and token billing against either TypeSafe or the Vercel AI Gateway.
Jev Integrations & Adapters#community#jevbest#catalog-import
69GitHub stars
Provider-neutral TypeScript runtime for System One decisions: write choice, noul, and score code once, then run it on TypeSafe Jev, local Reflex, any compatible /v1/systemone endpoint, or a custom provider.
Jev Integrations & Adapters#community#jevbest#catalog-import
34GitHub stars
Self-hosted agent runtime that adds an optional Jev judgment layer to memory filtering, lesson dismissal, request classification, and skill selection, with inspectable context and execution.
Jev Integrations & Adapters#community#jevbest#catalog-import
25GitHub stars
Agent runtime with a terminal UI and a messaging gateway that keeps memory, self-built skills, session search, scheduling, and subagents across Telegram, Discord, Slack, WhatsApp, Signal, and CLI.
Jev Integrations & Adapters#community#jevbest#catalog-import
51GitHub stars
GraphRAG layer that runs a 4-phase ingest-to-answer pipeline over Neo4j, Memgraph, AGE or Kuzu using Laya or Jev decisions instead of per-hop LLM calls.
Jev Integrations & Adapters#community#jevbest#catalog-import
5GitHub stars
Community Agent Zero plugin that exposes one typesafe_query tool for combined Choice, Noul, and Score questions over the same evidence, with probability cards in chat.
Jev Integrations & Adapters#community#jevbest#catalog-import
Back to all resources

From the Jev community

Posts from builders shipping with Jev right now.

Follow @typesafeai