{"skill":{"slug":"auto-evolving-agent","displayName":"auto-evolving-agent(智能体自主进化)","summary":"自我进化智能体，结合达尔文式探索与拉马克式优化，支持种群进化、交叉重组和多维适应度评估，用户可全程确认。","tags":{"latest":"1.0.0"},"stats":{"comments":0,"downloads":98,"installsAllTime":0,"installsCurrent":0,"stars":0,"versions":1},"createdAt":1776763873097,"updatedAt":1776764209815},"latestVersion":{"version":"1.0.0","createdAt":1776763873097,"changelog":"**Major overhaul: Skill redefined as a self-evolving agent using a hybrid evolutionary strategy, with previous orchestration/dispatch logic removed.**\n\n- Completely rewrote skill purpose and documentation to focus on hybrid evolutionary agent (Darwinian + Lamarckian) with population evolution, elitism, crossover, and multi-dimensional fitness scoring.\n- Removed all orchestration code, modules, and workflow scripts related to task dispatch and programming assistant; the skill is no longer a coordinator of sub-agents.\n- Added evolutionary strategy documentation, detailed process flows, scoring templates, and references (README.md, evolution-history-template.md, fitness-scoring-template.md, gene-pool-template.md).\n- New architecture details five evolution “rings” (layers), clearly outlining user roles, agent layers, gene pool, and reflective mechanisms.\n- Introduced user-in-the-loop, risk management via tiered confirmation, and guidelines for solution generation, ranking, and memory crystallization.\n- The skill is now a conceptual/design framework for evolutionary agents, not an execution orchestrator.","license":"MIT-0"},"metadata":null,"owner":{"handle":"zhuqingsonga","userId":"s17dmy3v6hzrhra9xgyyn3rxyd8412qa","displayName":"zhuqingsonga","image":"https://avatars.githubusercontent.com/u/33415526?v=4"},"moderation":{"isSuspicious":true,"isMalwareBlocked":false,"verdict":"suspicious","reasonCodes":["suspicious.llm_suspicious","suspicious.vt_suspicious"],"summary":"Detected: suspicious.llm_suspicious, suspicious.vt_suspicious","engineVersion":"v2.4.0","updatedAt":1776764209815}}