Best AI Agent AI SkillsPage 11

Use an orchestration pattern for decomposing, parallelizing, and reviewing Codex work.

A social platform client skill for AI agents.

Automate web browser interactions using natural language via CLI commands. Use when the user asks to browse websites, navigate web pages, extract data from websites, take screenshots, fill forms, click buttons, or interact with web applications.

Claw Brain provides personal AI memory, personality, bonding, and learning capabilities for OpenClaw/ClawDBot.

Autonomous financial research agent for stock analysis, financial statements, metrics, prices, SEC filings, and crypto data.

Proactive Agent is an architecture guide for an AI agent that anticipates needs, maintains memory, and improves over time, covering workspace memory, scheduled tasks, heartbeat checks, tool migration, and verification.

AI agent for deep research tasks

tg is a Telegram CLI guide for messages, chats and contacts.

Structure evidence-based analysis of complex decisions.

ima (ima.qq.com). Use this skill for ANY ima request — reading, creating, and updating data. Whenever a task involves ima, use this skill instead of calling the API directly.

Turns a novel or story idea into a structured short-drama prompt package delivered as JSON.

Opens a new topic from a message in a Telegram forum group.

Helps identify repetitive business processes, map bottlenecks and cross-system steps, design, document, implement, and monitor automations across sales, operations, finance, HR, and support, and assess expected value.

Drafts context-aware replies using the provided conversation and relationship background.

Plan an agent memory system combining vector retrieval, structured records, and file search.

Install and configure the local runtime core for NEXO Desktop.

Review an AI subscription stack for overlapping capabilities, usage-based scores and gap categories.

MCP Atlassian documents a Docker-hosted MCP server for Atlassian access.

Agent 上下文管理方法论：通过分层文件体系实现跨 session 记忆延续、职责分离和高效上下文恢复。Use when: (1) 搭建新 agent 工作区, (2) 优化 agent 记忆和上下文管理, (3) 长期运行 agent 的记忆维护。

A skill that handles cloud model rate limits and falls back to a local model.

Review source, files, and permission risks before installing or running an unfamiliar agent skill.

Google Forms API integration with managed OAuth. Create forms, inspect structure, review responses, change publishing settings, and manage response notificat...

Connect to Office 365/Outlook email, calendars, and contacts.

A local memory system for AI agents with persistent memory, affect tracking and semantic recall.

Skill Scaffold is a CLI for creating AI-agent skill structures.

Helps users discover and install agent skills when they ask questions like how do I do X. And also 50+ models for image generation, video generation, text-to...

Controls OpenCode over ACP: start sessions, send prompts, resume conversations.

A skill that backs up and restores ClawdBot configuration.

A skill for querying and controlling Home Assistant smart-home devices through MCP.

Plan coding, business analysis and read-only NAS metadata tasks.

Moltspaces lets agents join audio rooms to talk with other agents and users.

Moltpixel is a collaborative service where AI agents create and interact on a shared pixel canvas.

Serenity的思维框架与认知工具。基于5个维度的深度调研

quick-reminders documents one-shot message reminders for less than 2 days ahead.

Record reward and motivation state for agents and carry it into later sessions.

Extract structured facts from notes and transcripts, build a knowledge graph, and generate session briefings.

AI视频生产线的项目经理（第1环）：分镜脚本、素材交付清单、交付前字幕/音频/画面验收。五环流水线的交接物标准与衔接全部由它定义，中间不用手动倒文件。

CodeBuddy Code CLI installation, configuration and usage guide. CodeBuddy Code is Tencent's AI-powered CLI programming assistant supporting natural language driven development. - MANDATORY TRIGGERS: CodeBuddy, codebuddy, AI CLI, Tencent AI coding, @tencent-ai/codebuddy-code, terminal AI assistant - Use when: installing CodeBuddy CLI, configuring CodeBuddy, using CodeBuddy commands, troubleshooting CodeBuddy issues

ai-boss-assistant supplies executive-assistant persona and workflow templates.

Proactive Agent documents persistent-memory and proactive-work patterns.

WorkBuddy 的技能 / 专家 / 连接器统一管家，并充当「调度层」。四件事：①关闭基本不用的（有证据、带豁免、只关不删）②你描述需求，先从已装资产里挑出对口的建议加载 / 直接唤起 ③去 SkillHub 找更合适的、你没装就推荐安装 ④你给「需求 + 验收标准」，管家负责中间——路由已装 / 推荐未装 / 从 GitHub 调研造货并发布，最后按验收标准验证。

Write project documentation that AI agents can retrieve and use efficiently.

记忆文件根目录

面向没有产品设计或开发经验的小白用户，将“我想做一个工作台”之类的模糊想法，通过低门槛、多轮、动态的需求访谈，整理为包含目标、角色、场景、功能、页面、流程、数据、权限、异常、依赖、视觉和验收标准的完整工作台搭建提示词。当用户想制作工作台、整理工作台需求、优化已有工作台方案，或需要把业务想法交给工作台搭建师时使用。

A channel plugin that lets an agent exchange messages over WeCom.

在 SkillHub 平台查找/搜索 Skill 技能。基于 skills 列表接口，支持关键词分词搜索、一级标签（一级分类）筛选、以及二者组合检索。当用户说『找个 xxx 技能』『有没有处理 PDF 的 skill』『SkillHub 上搜一下 xxx』『按分类看技能』『办公效率类有哪些技能』『推荐一个做数据分析的 skill』『这个需求有现成技能吗』等需要在 SkillHub 上发现/检索/推荐 Skill 的场景时使用本技能。

phone-calls-bland uses Bland AI to make task-directed calls for the user.

Harness Engineering — AI Agent 复杂任务执行方法论 一个按需加载的 AI 编程 Agent 技能，让你的 Agent 在处理大型、多步骤软件任务时更可靠、更彻底、更持久。 核心能力： · 强制行为规范：每个功能完成后必须 commit + push，不允许偷懒、不允许半成品、不允许"看起来对了"就算验证通过 · 对抗性自审：Agent 在提交前自动切换为 Reviewer / QA / 架构师三种角色逐一审查，借鉴"门下省封驳"机制，不通过就打回重做 · 持续推进：TODO 列表驱动，不完成不停止，不跳步不切换，功能列表只允许改状态不允许删描述 · 分离评估：即使单 Agent 也能跑"生成 → 评估 → 迭代"循环，用运行时验证替代"我觉得没问题" · 上下文管理：子代理优先探索、按需读取、上下文焦虑检测、多会话交接协议，确保长任务不迷失 适用场景：跨 3+ 文件的重构、功能开发、性能优化、代码迁移、多会话大型项目。日常小改动无需加载。 兼容：Claude Code / Cursor / CodeBuddy / WorkBuddy / OpenClaw / OpenHarness 及所有支持 .md 技能文件的 Agent 平台。 知识来源：综合 Anthropic、OpenAI 等 15+ 个来源、近 60 篇文章的工程实践和 Claude Code 512K 行泄露源码分析。 GitHub: github.com/rickyma/harness-engineering

Habit formation and procedural learning for AI agents. Develop preferences and shortcuts through repetition. Part of the AI Brain series.

ralph-evolver develops improvement hypotheses from project context and tracks patterns and outcome trends across iterations.