self-improving 开源项目推荐（464 个）

aden-hive /hive

面向生产级 AI 的多智能体测试框架

EverMind-AI /Raven

框架之框架 • 为 RSI 打造：一个可信、持久、自进化的多智能体生态系统，用于全域协作。

xvirobotics /metabot

构建受监督的、自我进化的 Agent 组织的基础设施 | Infrastructure for supervised, self-improving agent organization. 在飞书/Telegram 手机端运行 Claude Code 或 Kimi Code（双引擎，两家原生订阅可直接使用），共享记忆、Agent 工厂、定时任务、通信总线。

lobehub /awesome-rsi

A curated research map of Recursive Self-Improvement (RSI): models, agents, harnesses, embodied systems, automated AI R&D, benchmarks, and safety.

A-EVO-Lab /a-evolve

The official repository of "Position: Agentic Evolution is the Path to Evolving LLMs".

sonichi /sutando

我的 AI Stand。白天实时运行，夜晚自我重写。召唤我的 AI 超能力。

AtomicBot-ai /atomic-hermes

The agent that grows with you

Protocol-zero-0 /evolution-kernel

Self-improving code for any git repo: give it a measurable goal, a coding agent (Claude Code or Aider) proposes changes, the kernel scores them, keeps winners and rolls back failures. Open-source AlphaEvolve / FunSearch-style evolution loop, sandboxed, with an auditable ledger.

deepelementlab /clawcode

ClawCode 是受 claude 代码启发的 python 和 Rust 实现，专注于代理和基于体验的演进，是一个开源编码代理 CLI，支持 Anthropic、OpenAI、Gemini、DeepSeek、GLM、Kimit、Ollama、Codex、GitHub 模型以及通过兼容 OpenAI 的 API 运行的 200+ 模型。

zou-group /sirius

SiriuS: Self-improving Multi-agent Systems via Bootstrapped Reasoning

kayba-ai /Kyoko

🔨 Kyoko is the all-in-one, fully local tool for debugging and improving your AI agents.

DomDemetz /claude-soul

Self-correcting learning engine for Claude Code — persistent identity, behavioral pattern tracking, and cross-session memory

OFA-Sys /DiverseEvol

Self-Evolved Diverse Data Sampling for Efficient Instruction Tuning

jjanczur /tyran

A task conductor for Claude Code that gets better the more you use it. Multi-agent orchestration with an evidence contract that BLOCKS instead of asking, gates written as hooks rather than prompts, and a retrospective that learns your repo's rules and tunes its own playbook.

ranausmanai /tinyforge

A tiny model that teaches itself to code better. On your laptop. No cloud. No teacher model. No human feedback.

griffinwork40 /agent-afk

在你的终端里开始跑一趟，然后离开。完成或需要你时收到提示。每一步都是可读的痕迹，发货前你会检查。

KishoreP1 /DetailCLIP

Detail-Oriented CLIP for Fine-Grained Tasks (ICLR SSI-FM 2025)

UniM0cha /self-improving-skills

Hermes Agent-style self-improvement for Claude Code

WhitehatD /crag

The bedrock layer for AI coding agents. One governance.md. Any project. Never stale. Universal skills + cross-agent compilation (Claude, Cursor, Codex, Gemini, Aider).

ruvnet /autogenous

Governed Evolutionary Software — Autogenous turns runtime failures into verified, portable, reversible software adaptations. AGL typed mutations, antibody packages, hard-gate promotion, automatic rollback.