What happened
promptfoo v0.121.16 (released 2026-06-16) adds native provider support for Anthropic's newly export-restricted Claude Fable and Mythos models and publishes all four promptfoo red-teaming skills to the Claude Code marketplace, enabling one-click red-team eval workflows directly from Claude Code.
Why it matters
Extends the most widely-used open-source LLM red-teaming framework to the newest Anthropic frontier models and embeds red-team capability into the Claude Code developer workflow, lowering the barrier for AI security testing at the point of development.
Applicability
Security engineers and AI developers using promptfoo for LLM red-teaming; Claude Code users who want in-IDE safety evaluation — evaluate now.