Self-refining AI agents for software development — critique every session, accumulate evidence, and automatically improve their own system prompts. Works with Claude Code & GitHub Copilot.
software-development ai-agents claude agent-framework self-improving github-copilot llm prompt-engineering anthropic context-engineering self-refining critique-loop
-
Updated
Mar 25, 2026 - Shell