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

openclaw-jev

Production control plane for OpenClaw agents: Jev routes each Slack message to its task, with Temporal workflows, an evaluator and memory.

Libraryproduction
View on GitHub ↗

Overview

OpenClaw + Jev is a production control plane, called the Reliability and Memory Plane, around OpenClaw agents. It runs Aura, a personal assistant on Slack. A plugin claims every DM; intake routing uses Jev to decide whether a message joins running work, starts new work or a follow-up, or is unclear, with an LLM analyst for cases Jev is unsure about and a single clarifying question to the user when still ambiguous.

Work runs as durable Temporal workflows, a separate Process Evaluator session judges results before anything reaches the user and can send work back for rework, replies are posted idempotently in ordered parts, and memory is stored in Postgres with a Qdrant index for deep recall during and after answers. It addresses gaps such as follow-ups starting new conversations, done meaning only that the model said so, and lost or unchecked replies after crashes. Built by Hyper-AI-Lab.

What the decision model decides

TypeQuestionAnswers
ChoiceDoes this message join running work, start new work, or need clarification?(inferred)join · new · follow-up · unsure

How it works

  1. Plugin claims the Slack message.
  2. Jev classifies it at intake; an LLM analyst handles uncertain cases.
  3. Temporal workflow runs the task in OpenClaw; an evaluator accepts or reworks.
  4. Reply is posted idempotently and memory updated.

Strengths

  • Follow-ups join the right task.
  • Results are checked before delivery.
  • Durable workflows survive crashes.

Caveats

  • Complex stack: OpenClaw, Temporal, Postgres, Qdrant, Slack.
  • Requires a TypeSafe key for routing.

At a glance

Type
Library
Status
production
Language
Python
Runtime
Temporal workers, OpenClaw plugin
Has code
yes
Has tests
yes

Use cases and integrations

How curated lists describe it

Frequently asked questions

What does Jev decide?

Where each incoming Slack message belongs: existing work, new work or clarification.

What if Jev is unsure?

An LLM analyst takes over, and if still unclear the user gets one question.

Is it used in production?

Yes, it runs the Aura assistant on Slack.

Related projects

  • Intent-Router

    Intent compiler skill that turns vague coding requests into a typed IntentSpec, with Jev or LAYA as optional decision backends.

    Projectbeta
  • Jev by Example

    Ten runnable JavaScript lessons where Jev judges agent situations like memory conflicts and handoffs while code enforces explicit policy.

    DemoprototypeJavaScript
  • jev-agent-skills

    Portable skill pack giving agents a conservative Jev second opinion on intent, completion, routing and bounded browser navigation.

    ProjectbetaPython
  • jev-chat-windows★ 802

    Windows WeChat reply assistant: local offline OCR reads the chat, Jev judges intent and tension and ranks three drafted replies to fill in.

    ProjectproductionPython
  • astrbot_plugin_jev_active_reply

    AstrBot plugin where Jev decides whether the bot should join a group conversation; the main model still writes the reply.

    ProjectproductionPython
  • fast-jev-compaction★ 7,616

    Claude Code plugin and npm library that replaces the compaction summary with Jev keep, truncate or drop scores for every tool call.

    PluginbetaTypeScript