Promptfoo

Overview

Promptfoo is an open-source framework for testing LLM prompts — define test cases, run them against one or more models, and assert on outputs. Think "unit tests for prompts." Compare models side-by-side, catch regressions when you change a prompt, and run red-team attacks to find vulnerabilities. Web UI for viewing results, CLI for CI integration.

When to Use

• Changing a prompt and want to make sure it doesn't break existing behavior
• Comparing model performance (GPT-4o vs Claude vs Gemini on your use case)
• Red-teaming an LLM application for prompt injection and harmful outputs
• Building a prompt evaluation suite for CI/CD
• Systematic prompt engineering (not vibes-based)

Instructions

Setup