Agent Memory
AI agents forget everything between sessions. This skill builds persistent memory systems — from simple filebased approaches to full vectorsearch architectures — so agents retain context, learn from past interactions, and make better decisions over time.
概览
The Agent Memory skill, part of the TerminalSkills/skills repository, addresses the common limitation where AI agents lose context between individual sessions. By implementing this skill, developers can establish persistent memory systems ranging from basic file-based storage to advanced vector search architectures. This infrastructure allows agents to retain historical interaction data, learn from previous user exchanges, and improve decision-making accuracy over time. Compatible with platforms like Claude, Gemini, and Cursor, the skill provides a framework for building more intelligent, context-aware autonomous systems. The repository, which has earned 71 stars, offers these capabilities through TypeScript and Python implementations, ensuring that agents maintain a continuous knowledge base rather than starting from a blank slate during every new session.
使用场景
安装说明
# Review source first
open https://github.com/TerminalSkills/skills/blob/main/skills/agent-memory/SKILL.mdCopy or clone the skill folder into your agent skills directory after reviewing its instructions and scripts.
安全提示
Implementing persistent memory requires careful management of stored interaction data to prevent unauthorized access to sensitive session history. Users should ensure that file-based or vector databases are properly secured and that any data retained by the agent complies with local privacy standards and organizational data retention policies.
相关 Skills
Constant Time Testing
trailofbits/skills
Constant-time testing 用于检测加密代码中的定时侧信道。在审计加密实现的定时漏洞时使用。
Managed Deep Agents
langchain-ai/langchain-skills
在使用 mda CLI 在 LangSmith 中构建、测试或部署 Managed Deep Agents 时，INVOKE THIS SKILL。涵盖了代码优先、基于文件的项目布局；define_deep_agent / defineDeepAgent；编写的 tools 和 middleware；MCP 连接器；cron 计划任务；skills；sandboxes；mda init/dev/deploy；Context Hub；以及 Python 和 TypeScript 中的人工在环（human-in-the-loop）中断。
GSAP Core
mxyhi/ok-skills
核心 API 的官方 GSAP skill —— 包含 gsap.to()、from()、fromTo()、easing、duration、stagger、defaults、gsap.matchMedia()（响应式、prefers-reduced-motion）。当用户请求 JavaScript 动画库、React/Vue/vanilla 中的动画、GSAP 补间动画、缓动、基础动画、响应式或减弱动态效果 a 时使用。
花叔Design · Huashu-Design
mxyhi/ok-skills
花叔Design——用HTML做高保真原型、交互Demo、幻灯片、动画、设计变体探索+设计方向顾问+专家评审。根据任务embody对应专家（UX/动画师/幻灯片设计师/原型师），避免web design tropes。触发词：做原型、交互原型、HTML演示、动画Demo、设计变体、hi-fi设计、UI mockup、prototype、做个HTML页面、做个可视化、app原型、iOS原型、导出MP4/GIF、60fps视频、设计风格、设计方向、配色方案、推荐风格、选个风格、做个好看的、评审、好不好看、review this design、带解说的动画、解说视频、长视频科普、voiceover、narration、5分钟讲清楚什么是X
Design Claude
tinh2/skills-hub-registry
以 HTML 为中心的高保真设计技能：交互式原型、幻灯片、动画演示、设计变体探索、视觉方向建议和专家评审。触发词：制作原型、设计演示、交互式原型、HTML 演示文稿、动画演示、设计变体、高保真设计、UI 模型、原型、设计探索、制作 HTML 页面、制作可视化、应用原型、iOS 原型、移动应用模型、导出 MP4、导出 GIF、60fps 视频、设计风格、设计方向、设计 p
Design Build
tinh2/skills-hub-registry
以 Ralph Wiggum 的速度构建独特的生产级界面。交付可运行的代码（Web 或移动端），包含现代 CSS、流体排版、有意义的动效，且无通用的 AI 审美。适用场景：'build UI'、'design this'、'create interface'、'make it look good'、'build screen'、'implement design'、'frontend build'、'design build'。