Open-Source LLM Testing, Red Teaming and Evaluation

Promptfoo is an adversarial testing platform that helps developers harden AI applications against jailbreaks, prompt injection, and unsafe model behavior. Widely adopted across industry — including by more than 150,000 developers — Promptfoo generates context-specific attacks rather than relying on static prompt lists, enabling realistic and application-aware red teaming.

The platform supports multi-model evaluation, integrates directly into CI/CD pipelines, and maps security findings to frameworks such as OWASP, NIST, MITRE ATLAS, and the EU AI Act.

Promptfoo secured a $18.4 million Series A from Insight Partners with participation from Andreessen Horowitz and is among the most widely used opensource foundations for production-grade LLM security testing.