guardrail-chatbot-jev
Python/TypeScript/Go chatbot safety library checking user input and bot replies with Jev and returning one actionable verdict per message.
Overview
guardrail-chatbot-jev is a content-safety library for AI chatbots: it checks what the user sends and what the bot replies and returns one clear decision to act on, such as allow, hold, mask a phone number in the reply, or send a person to a crisis line. It targets cases like attempts to extract dangerous instructions, pasted personal ID numbers, and confident medical advice to someone in distress.
It is a library rather than a service. Python and TypeScript packages read the same policy file so both sides of a stack cannot drift, and neither has third-party dependencies; a Go SDK offers the same engine with live, replay and regression tools. Features include three checks, multi-turn attribution, realtime review, caching, a prefilter, sessions, streaming, offline tuning and a CLI. A Viet Nam compliance policy with prewritten replies in Vietnamese, English and Chinese is available, and a live demo runs on a Vietnamese assistant. Releases are licensed CC BY-NC 4.0, which restricts commercial use.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Choice | What should happen with this message: allow, hold, mask, or redirect to crisis support?(inferred) | — |
| Yes / no | Does this message contain personal data that should be masked?(inferred) | yes · no |
How it works
- Check user input before it reaches the model
- Check the bot reply before it reaches the user
- Jev judgments mapped to a verdict by the shared policy file
- Your code acts on the verdict
Strengths
- Same policy across Python, TypeScript and Go
- No third-party dependencies
- Regional compliance policy available
Caveats
- CC BY-NC 4.0 licence restricts commercial use
- Requires a TypeSafe key
Getting started
Install the Python, TypeScript or Go SDK from the tagged releases and load a policy file.At a glance
- Type
- Library
- Status
- production
- Language
- Python
- Runtime
- library (Python, TypeScript, Go)
- Has code
- yes
- Has tests
- yes
- License
- CC-BY-NC-4.0
Use cases and integrations
How curated lists describe it
It is a library, not a service. You call it, you get a verdict, and your code decides what to do. It runs in \*\*Python and TypeScript\*\*, both reading the same policy file, so the two sides of your stack cannot drift apart. Neither package has a third-party dependency.
logicrw/awesome-jev-projects
Frequently asked questions
Is it a hosted service?
No, it is a library your code calls.
Which languages?
Python and TypeScript sharing one policy file, plus a Go SDK.
Can I use it commercially?
The releases are CC BY-NC 4.0, which restricts commercial use.
Related projects
Gateway admission control that asks Jev what each incoming request is trying to do, filtering prompt injection and abuse.
AppbetaLuaSystem 1 ingestion gate that asks Jev intent, risk and sensitivity in one call, then blocks, answers deterministically or passes to an LLM.
ProjectprototypePythonPi guard plugin with timing-based rules, Jev risk checks on tool calls, output redaction and reminders for repeated failures.
PluginprototypeTypeScriptGo reverse proxy between nginx and an LLM backend that extracts each request's user input, asks Jev if it is harmful, and blocks or passes.
ProxybetaGoOne-file Mastra input processor that asks Jev whether a message must be blocked, failing open behind a deadline and circuit breaker.
ProjectproductionTypeScriptBenchmark testing whether Jev or open Kev can replace an LLM judge in agent guardrail callbacks on 7,925 public examples.
ProjectresearchPython