Autonomous coding is the capability of an AI agent to write, modify, test and debug software with minimal human intervention, planning and executing multi-step coding tasks end to end. It builds on code generation but adds agentic control loops for tool use, self-correction and iterative verification against a goal. It is a core capability enabled by agent frameworks, orchestrators and execution sandboxes.