companies
Promptfoo
Makes the open-source CLI developers use to test prompts, red-team LLM apps for jailbreaks and injection, and catch quality regressions before shipping.
Ian Webster and Michael D'Angelo started Promptfoo in 2024 after Webster ran into a recurring problem building LLM features at Discord: prompts that worked fine in a demo would quietly break in production, with no good way to test them systematically. Promptfoo began as a command-line tool for running the same set of test cases against different prompts or models side by side, closer to a testing framework than a dashboard. The project's red-teaming features, which scan applications for jailbreaks, prompt injection, PII leakage, and other vulnerabilities using dozens of built-in attack techniques, turned it into one of the more widely used AI security tools; by early 2026 it counted more than 300,000 developers and adoption at over a quarter of the Fortune 500, including internal use at OpenAI and Anthropic. The project stayed MIT-licensed throughout. In March 2026, OpenAI announced it was acquiring Promptfoo to fold its testing and red-teaming technology into OpenAI's own agent products, while committing to keep the tool open source and continue supporting existing customers.
- Founded
- 2024
- Headquarters
- San Francisco, United States
- Sector
- infrastructure
Tools
- Promptfoo
Promptfoo evaluates and red-teams LLM apps, prompts, models, agents, and RAG systems through a CLI, library, and web interface.