I Built a 7-Agent Prompt Framework, Then Used It to Debug Its Own Output

The author developed a 7-agent prompt framework named C.E.H. running on local LLMs, which successfully built a complex RAG system. Faced with 14 failures in the generated code, the author uniquely used the C.E.H. framework itself to debug and fix its own output.