NOPE Documentation

NOPE is a safety layer for chat applications and LLMs. Detect user mental health crises, analyze AI assistant behavior for psychological harm, verify system prompt compliance, and match users to crisis resources from 222 countries.

Products

Evaluate

$0.003/call

Edge-backed safety assessment. 9 risk types with severity, resources, and reasoning.

Oversight

Limited

AI behavior analysis. Detects 85 harmful patterns like gaslighting and boundary violations.

Steer

$0.001/call

System prompt compliance. Verify AI responses follow your rules.

Signpost

Free / $0.001

Crisis helplines from 222 countries. LLM-ranked matching optional.

Quickstart

Get your API key and make your first call in 5 minutes.

API Reference

Complete reference for all endpoints, types, and response fields.

Taxonomy

Risk subjects, types, features, severity scales, and protective factors.

API Sandbox

Try the API interactively in your browser.

Core Concepts

Subject × Type Classification

NOPE uses an orthogonal design separating WHO is at risk from WHAT the risk is:

Subject (who)

self other unknown

Type (what)

suicide abuse violence +6 more

e.g., "My friend is suicidal" → subject: other, type: suicide

Severity + Imminence

Each assessment includes both severity (how serious: none → critical) and imminence (how soon: chronic → emergency). This lets you calibrate your response—a "high severity, chronic" situation needs different handling than "moderate severity, emergency."

Crisis Resource Matching

Beyond detection, NOPE intelligently matches crisis resources from a curated database of verified helplines across 222 countries. Resources are scored and ranked by relevance to the specific risk detected—a youth eating disorder hotline ranks above a generic crisis line when that's what's needed.

AI Behavior Taxonomy

Oversight analyzes AI assistant responses for 85 harmful behaviors across 14 categories:

gaslighting love_bombing boundary_violation crisis_minimization +81 more

Detects psychological manipulation, inappropriate relationships, and safety failures in AI responses.

Guides

Integration Patterns

Recipes for chat, background evaluation, and non-chat contexts

Evaluation API

Edge-backed safety assessment with SB243 compliance

System Prompt Compliance

Verify AI responses follow your rules

AI Behavior Oversight Limited Access

Analyze AI assistant conversations for psychological safety

Signpost

Understand and render matched helplines

Widget Integration

Embed pre-built crisis resource UI

SDKs

Python

pip install nope-net

PyPI · GitHub

Node.js

npm install @nope-net/sdk

npm · GitHub

Understanding NOPE

Methodology

Architecture, clinical grounding, limitations, and what NOPE doesn't claim.

Test Suites

Public test results for external verification and accountability.

Need help?

Check status.nope.net for API status, or email [email protected] for support.