promptfoo-redteam-llmlisted

# Red Team an LLM Application with Promptfoo Run the canonical promptfoo red team workflow against an LLM application: configure prompts and targets, generate adversarial test cases via plugins, wrap them in attack strategies, run the pentest, and review the vulnerability report. ## Use this skill when - Setting up the first red team for an LLM application - Configuring `promptfooconfig.yaml` (prompts, targets, plugins, strategies, grader) - Generating adversarial tests with `promptfoo redteam generate` / `redteam run` - Reviewing the vulnerability report with `promptfoo redteam report` - Scoping a red team to specific OWASP LLM Top 10 categories (LLM01, LLM08, LLM09) - Comparing multiple LLM targets side-by-side ## Do not use this skill when - Testing a content filter / guardrail service directly → use `promptfoo-redteam-guardrails` - Testing a RAG system with retrieved context → use `promptfoo-redteam-rag` - Testing LLM agents with tools/state/memory → use `promptfoo-redteam-agents` - Testing an app with multiple input fields (user_id + message) → use `promptfoo-redteam-multi-input` - Testing vision/audio/video models → use `promptfoo-redteam-multimodal` - Assessing a foundation/base model in isolation → use `promptfoo-redteam-foundation-models` - Setting up CI/CD drift detection or supply chain gates → use `promptfoo-redteam-supply-chain` ## 🛡️ Edge cases (mandatory handling) Every red team config MUST handle these defensively — not just the happy path: - **Missing gr