Skip to main content

​
Overview

Merit uses environment variables for different features. Here’s what you need for each:
FeatureRequired VariablesGet Access
Basic TestingNoneBuilt-in
AI PredicatesMERIT_API_KEYContact us
Error AnalyzerYour own LLM keysDirect from provider

​
AI Predicates (LLM-as-a-Judge)

AI predicates (has_facts, has_topics, etc.) call the Merit cloud service.

​
Required

.env
MERIT_API_BASE_URL=https://api.appmerit.com
MERIT_API_KEY=your_merit_api_key_here

​
Optional

.env
# Enable debugging mode (includes reasoning in responses)
MERIT_DEBUGGING_MODE=true

# Advanced: Customize connection pooling
MERIT_API_KEEPALIVE_EXPIRY=30

​
Get API Access

Contact us to get your Merit API key.

​
Error Analyzer (Bring Your Own Key)

Merit Analyzer uses your LLM provider credentials directly. Currently supports Anthropic models via direct API or AWS Bedrock.

​
Anthropic (Direct)

.env
ANTHROPIC_API_KEY=sk-ant-...
MODEL_VENDOR=anthropic
INFERENCE_VENDOR=anthropic
Get your key: console.anthropic.com

​
AWS Bedrock (Anthropic)

.env
MODEL_VENDOR=anthropic
INFERENCE_VENDOR=aws
AWS_ACCESS_KEY_ID=AKIA...
AWS_SECRET_ACCESS_KEY=...
AWS_REGION=us-east-1
Configure AWS credentials: AWS Documentation

​
Loading Environment Variables

Merit automatically loads .env files from:
  1. Current working directory
  2. Project root (if different)

​
.env File

Create a .env file in your project root:
.env
# AI Predicates
MERIT_API_BASE_URL=https://api.appmerit.com
MERIT_API_KEY=your_merit_api_key_here

# Error Analyzer (choose one provider)
ANTHROPIC_API_KEY=sk-ant-...
MODEL_VENDOR=anthropic
INFERENCE_VENDOR=anthropic

​
Shell Export

Or export directly in your shell: 
export MERIT_API_KEY=your_merit_api_key_here
export MERIT_API_BASE_URL=https://api.appmerit.com

​
CI/CD

Store as secrets in your CI/CD platform: GitHub Actions: 
env:
  MERIT_API_KEY: ${{ secrets.MERIT_API_KEY }}
  MERIT_API_BASE_URL: https://api.appmerit.com
GitLab CI: 
variables:
  MERIT_API_BASE_URL: "https://api.appmerit.com"
  MERIT_API_KEY: $MERIT_API_KEY  # From CI/CD variables

​
Quick Reference

​
I want to use AI predicates in my tests

.env
MERIT_API_BASE_URL=https://api.appmerit.com
MERIT_API_KEY=your_merit_api_key_here
Get API key →

​
I want to run the error analyzer

ANTHROPIC_API_KEY=sk-ant-...
MODEL_VENDOR=anthropic
INFERENCE_VENDOR=anthropic

​
I want both features

.env
# For AI predicates
MERIT_API_BASE_URL=https://api.appmerit.com
MERIT_API_KEY=your_merit_api_key_here

# For error analyzer
ANTHROPIC_API_KEY=sk-ant-...
MODEL_VENDOR=anthropic
INFERENCE_VENDOR=anthropic

​
Troubleshooting

​
”MERIT_API_KEY not configured”

You need Merit API credentials to use AI predicates:
  1. Contact us to get an API key
  2. Add to .env file: 
    MERIT_API_KEY=your_key_here
MERIT_API_BASE_URL=https://api.appmerit.com

​
“API key not configured” (Analyzer)

The analyzer needs LLM provider credentials:
  1. Get API key from your provider (OpenAI, Anthropic, etc.)
  2. Add to .env file: 
    OPENAI_API_KEY=sk-...
MODEL_VENDOR=openai
INFERENCE_VENDOR=openai

​
Variables not loading

  • ✅ Check .env file is in project root
  • ✅ Check file is named exactly .env (not .env.txt)
  • ✅ Restart terminal/IDE after creating .env
  • ✅ Check for typos in variable names

​
Security Best Practices

Never commit API keys to version control!

​
Add to .gitignore

.gitignore
.env
.env.local
.env.*.local

​
Use Different Keys Per Environment

# Development
.env

# Production
.env.production

# CI/CD
# Store in CI/CD secrets, not in files

​
Rotate Keys Regularly

  • Change keys periodically
  • Revoke compromised keys immediately
  • Use separate keys for dev/staging/production

​
Next Steps

Quick Start

Get started with Merit in 2 minutes

AI Predicates

Learn about LLM-as-a-Judge assertions