Skill Sedimentation · 经验沉积：把单次任务收成可复用知识
Skill Sedimentation · 经验沉积是 SkillHub 社区技能（@user_310a65be/skill-sedimentation，v3.0.1，ai-agent），通过 Trigger→Retrieve→Act→Reflect→Distill→Promote 将单次任务经历转为可复用知识资产，且 requires_api_key=false。本文依据 SkillHub 页面与 skillhub.md 说明安装 CLI、用 --dir 安装到 Cursor 等 skills 目录，并说明适用场景与以 SKILL.md 为准的使用边界。
Read Moreauto-evolving-agent：用进化论思路组织智能体自我改进
auto-evolving-agent（@user_dfbdc5ab/auto-evolving-agent，v1.0.0）是 SkillHub 社区 ai-agent 类 Skill，用达尔文探索与拉马克优化相结合的混合策略，组织种群候选、交叉重组、六维适应度评分与竞技场排名，并在五环架构下通过 Git 快照、审计与分级确认保证用户主导。本文基于官方页面与 SKILL.md 介绍定位、核心能力、SkillHub 安装命令（含 Cursor 等 skills 目录）及中英文触发语，并说明适用 OpenClaw 开发者及进化改文件时的安全注意。
Read MoreSelf-Evolution 自我进化：为 AI Agent 加入自我审视与持续进化
Self-Evolution 是 SkillHub 社区 Skill（v1.0.0，ai-agent），官方简介强调让 Agent 具备自我审视与持续进化能力，并对标 Hermes Agent。标签 requires_api_key=false。本文依据 SkillHub 页面与安装文档说明：通过 skillhub CLI 或安装提示词将 @user_5953718f/self-evolution-v2 装入 Cursor、Claude Code、Codex 等 skills 目录，并说明适用场景与公开资料未覆盖的细节需以 SKILL.md 为准。
Read MoreDeepSeek Harness Plugin: satan9394/dsh-self-improving-agent
The DeepSeek Harness (DSH) plugin, dsh-self-improving-agent, addresses challenges such as continual agent evolution, cross-session memory, and automated task handling. Through code integration, it constructs a self-evolving framework containing seven core capabilities, including closed-loop learning, automatic skill creation and improvement, and cross-session recall. Since no official installation command is currently available and the documentation is incomplete, users need to obtain the plugin from the GitHub source code. It is suitable for complex agent scenarios requiring long-term operation and self-optimization, but users should be mindful of permission risks and verify source-code compatibility.
Read More