I built a load tester with an AI diagnosis layer—because no existing tool does both

QueryScope combines load testing (like k6/JMeter) with LLM-powered diagnostics by embedding benchmark results via OpenAI and indexing them in Azure AI Search. A RAG pipeline using LangChain retrieves semantically similar and recent runs to generate grounded explanations of performance issues. It also includes an MCP server that lets Claude Desktop autonomously run benchmarks and analyze results.