Jev 101
Connect your bot
Search
中文 EN

use cases verify-llm-output

safety ✦ 82~40 min setupintermediate

Verify an LLM answer before show

Placeholder guardrail. Launch theme: verify / guardrail / jailbreak detection. Outline only.

Source: TypeSafe launch ↗ ✓ verified 2026-09-19
1

Copy the prompt below.

2

Paste it into Grok — open your Jev and send it as your first message.

3

Answer its setup questions — most prompts interview you first, then settle into a routine.

The prompt

Prompt 
OUTLINE ONLY — Not a production recipe.

When to use: LLM draft must pass a gate.
Inputs: draft + policy text as state.
Question sketch:
- policy_ok: Boolean
- jailbreak_suspected: Boolean
- quality: Score on your rubric
Confidence policy: fail closed on low confidence.
Links: TypeSafe launch “verify everything”.

Needs access to: typesafellm

How it's set up

LLM writes; Jev judges; code branches. Do not ask Jev to rewrite prose.

Not a production recipe.

Why it works

Keeps generation and judgment in different model classes.

This use case was collected from the public Jev community and is credited to its original author. Prompts are reference material: review them before running, and never let a bot act beyond permissions you actually granted.