Benchmark performance of RAG and RLM-based MCP servers for 1C codebases using automated E2E testing to optimize AI assistant integration.