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

uipath-jev-guardrail-connector

UiPath bring-your-own-guardrail connector that runs AI Trust Layer llm_as_judge checks as Jev noul questions.

Projectprototype
View on GitHub ↗

Overview

This unofficial demo is a UiPath Integration Service connector that lets UiPath's AI Trust Layer run its llm_as_judge guardrail against Jev. You write a policy in plain language, and every agent prompt, tool call and LLM response is judged against it in a single sub-second model call. The author notes that a judge guardrail is a yes/no question with a threshold, exactly a Jev noul, and that one request can carry a whole battery of policies.

The bring-your-own-guardrail contract needs two activities: GET /validators, answered locally by a hook so no call leaves UiPath, and POST /evaluate, which returns PASSED, FAILED or ERROR. Hooks are the source of truth and a local harness runs them for testing. It requires a UiPath tenant with BYOG enabled, an org admin, the uip CLI, Node 18+ and a TypeSafe key.

What the decision model decides

TypeQuestionAnswers
Yes / noDoes this prompt, tool call or response violate the written policy?PASSED · FAILED

How it works

  1. UiPath calls the connector for a guardrail check.
  2. The connector sends the input and policy to Jev as a noul.
  3. The probability is thresholded.
  4. The connector returns PASSED, FAILED or ERROR.

Strengths

  • Sub-second policy checks inside UiPath.
  • Several policies in one request.
  • Local test harness for hooks.

Caveats

  • Unofficial demo, not supported by UiPath or TypeSafe.
  • Requires UiPath BYOG access and org-admin rights.

At a glance

Type
Project
Status
prototype
Language
JavaScript
Runtime
UiPath Integration Service connector
Has code
yes
Has tests
yes

Use cases and integrations

How curated lists describe it

Frequently asked questions

What does Jev decide?

Whether an agent prompt, tool call or response violates a plain-language policy.

Is it supported by UiPath?

No, it is an unofficial demo.

What does UiPath receive?

PASSED, FAILED or ERROR from the evaluate activity.

Related projects

  • tripwire

    AI SDK middleware and OpenAI-compatible proxy that runs seven Jev checks on every LLM response, gated by confidence.

    ProxyprototypeTypeScript
  • jes

    Open-source agent guardrails that use Jev or LAYA to check prompts, retrieved content, tool calls and outputs for injection and risk.

    ProjectbetaPython
  • jev-enforce

    Claude Code plugin that checks every reply and file edit against your CLAUDE.md rules with one Jev yes/no per rule.

    PluginbetaTypeScript
  • jev-gates

    Claude Code plugin with nine Jev-judged gates for rules, scope, intent, completion and claims that escalate but never approve.

    HookbetaTypeScript
  • Jev4Mellea

    Unofficial Python adapter that brings Jev verification, classification and scoring into the Mellea framework as Requirements.

    ProjectbetaPython
  • jevshield

    Python pre-check layer that uses Jev Choice/Noul/Score to route agent requests to roles and gate side-effecting tool calls.

    ProjectbetaPython