Awesome-Self-Evolving-Coding-Agents
A curated survey repository of research, benchmarks and products on coding agents that improve themselves from execution outcomes, memory, tools and accumulated experience.
This repository accompanies a survey on self-evolving coding agents: coding agents that learn from execution outcomes, trajectories and environmental feedback to improve the persistent parts of their own workflow. It organizes such systems into three layers: assets (memory, skills, tools, context), architecture (harness, workflow, multi-agent structure), and model weights.
It curates five groups of resources: self-evolving coding agent systems organized by which layer they modify, general self-evolution methods evaluated on coding tasks, benchmarks and empirical studies of repository-level and general coding agents, deployed products with persistent adaptation mechanisms, and related surveys on self-evolving and coding agents.
It is a reference and paper list rather than a runnable tool.