Open-source toolkit for evaluating, comparing, and red-teaming LLM prompts and applications. By 2026, Promptfoo is a popular CI-first option for prompt regressi
Promptfoo is the developer-native eval and red-team toolkit: declarative test suites for prompts/models/RAG, CI integration, and automated adversarial probes (injection, jailbreaks), open-source, local-first, and the fastest way to make LLM quality a build gate.
Pricing: Open-source free; enterprise platform adds team features and managed scanning (published pricing, mid-2026).
Promptfoo excels at pre-deploy gating in CI (assertions, matrices, red-team); LangSmith at production tracing and dataset workflows. Many teams run both: gate with promptfoo, observe with a tracer.
Yes: its red-team mode generates adversarial inputs (injection, jailbreak, PII probes) against your app config, turning LLM security testing into a repeatable scan.