Code for "Evolutionary System Prompt Learning for Reinforcement Learning in LLMs" (arxiv.org/abs/2602.14697)
Offline prompt evolution engine for multi-agent systems, benchmark suites, and inspectable prompt DNA.
Multi-agent orchestration with self-evolution, cognitive memory, and governance.
面向生产环境的 AI Agent 动态技能网关与自进化底座：三层级联意图路由(R@1=98%)、8维沙箱评估、8重防御防线、LangGraph状态图旁路与SQLite原子发布状态机。
🧬 Stop guessing which LLM prompt works best. ab_explorer evolves prompts via automated genetic A/B testing: it breeds, crosses, and mutates candidates, scores them against your rubric, and selects the fittest across generations. You get a winner balanced for accuracy, cost, and latency — with a beautiful CLI that reports every step with early stop
HyperAgents-based framework for evolutionary prompt optimization using task agents, evaluators and meta-agents.
Endgame-AI - Local desktop automation agent powered by LLMs. Control your computer with natural language using local models.
Breed prompts as collectible creatures, then prove their traits in blind, evidence-based trials.
Polish a raw idea with facts from web search — and evolve the instructions that drive the polishing
To associate your repository with the prompt-evolution topic, visit your repo's landing page and select "manage topics."