A comprehensive testing toolkit that gives Claude access to over 130 specialized QA and development tools focused on AI safety and evaluation. It exposes operations for testing prompt injection vulnerabilities, evaluating RAG systems, benchmarking vision language models, and validating guardrails. You'd reach for this when you need to systematically probe AI applications for weaknesses, run structured evaluations against your agent pipelines, or automate quality checks that typically require manual testing. The breadth here means you can test multiple AI components from a single integration, from basic prompt fuzzing to multimodal output validation. Runs via streamable HTTP and is completely free to use.
claude mcp add --transport http io.github.jcjamet-ia-qa-toolbox https://www.ia-qa.com/mcp