1,266 boosters for "claude-code" — open source, verified from GitHub, ready to install
基于 Benjamin Bloom「2 Sigma Problem」研究（1984）的一对一 AI 导师系统。每个课题是一个独立文件夹，通过自适应生成的课程文档 + 用户反馈循环模拟一对一苏格拉底式导师，把学习效果推向 +2σ。学习的主要载体是文档，对话只是辅助确认状态。 所有回复、解释、提问、文档一律使用中文。 触发本 skill 后，以下守则在整个学习交互全程生效——违反字面就是违反精神：
用户在学 / 接触一个新概念 X（新技术、新算法、新理论、新领域……），尤其觉得"陌生 / 有点难"的时候。难，往往不是智商问题，是它相对用户还存在"没接上的旧知识"。 用一两句话说清 X 到底在干什么——它的核心机制 / 结构是什么。剥掉术语外壳，留下"它本质是一个 "。只有先拿到结构，才能去匹配用户学过的东西。 拿不准就直接问「你学过 吗？」，绝不从正在讲的材料 / 文章作者背景推断用户会什么。
用户说"想学 / 理解 / 搞懂 / 讲讲一个概念 X"时——这是默认入口，一次跑完五视角，用户再选深入哪个。 抓住 X 的本质结构（剥术语），按三猜想给 🎁其实已学过 / 🔗结构同构（字段级对应表）/ 🧩可用已有知识解释，点出元知识。先激发信心，再谈深入。 定位"既定问题"（学 X 解决什么）、现有知识够不够、X 的贬值速度与 ROI，给"够用就停 / 只学最小那块 / 值得深挖"的深度边界。不是劝退，是防止一上来过度钻。
用户学完一个东西想验真伪，或隐约觉得"好像懂了但不踏实"。也是"重输入轻输出"的一次强制输出。 请他用自己的话、把你当外行，把概念讲一遍。别让他背定义——要他解释、打比方。 专挑他含糊带过、用术语糊弄、跳过的环节追问："为什么？""那这个是怎么来的？""举个例子？"命中他答不上来或开始绕的地方。
用户要动手做 / 研究一个东西，或想把某个已有产出改得更好。这是"重输入、轻输出"短板的解药——逼用户从输入切到输出。 别追求完美，先有一个能跑 / 能看的最小版本。卡在"还没准备好"就是没进改良主义。 针对缺陷提一个改良策略（视为假说，可对可错），动手改，看效果。错了也有用——错误暴露后，下次自动规避这个方向。
"name": "design-council", "description": "Convene parallel role-specialized peer agents (dynamic roster, plan card first) to debate a cross-domain decision or audit a codebase in real time. Invoking Claude acts as CEO: convenes, routes peer-DMs, arbitrates deadlocks, writes a one-page decision log."
Create detailed implementation plan using the implementation-planner agent. ⚠️ Requires ResearchPack from or 1. Checks ResearchPack exists and passed validation (80+)
"name": "prompt-engineering-skills", "description": "아이디어를 실행 가능한 프롬프트로 정리해주는 도구 — /prompt v3.3.0 + 모델별 최적화 전략(Claude/GPT/Gemini/이미지) 스킬 + 공냥 프롬프트 킷 v4 동봉(이미지 정본)", "name": "tofukyung",
"description": "Skills for working with the Thermite SIMD library.", "source": "./.claude/skills/thermite", "description": "Usage guide for the Thermite generic SIMD library and its companion crates (special functions, autodiff Dual, Compensated double-double, SDF, geometry, FFI).",
"$schema": "https://anthropic.com/claude-code/marketplace.schema.json", "name": "cross-request-marketplace", "name": "leeguooooo",
A fully autonomous AI research agent that ingests sources into Google NotebookLM, runs deep web research, synthesizes knowledge through cited Q&A and 9 downloadable artifact types, creates polished content drafts, and optionally publishes to social platforms.
"name": "brave-search", "name": "Brave Software" "name": "brave-search-skills",
Auto-loaded when writing documentation, content, or user-facing text. Enforces consistent, direct voice and eliminates LLM rhetoric. Also loaded when a project sets or in , to govern conversational response length and structure (see Output Modes). Bad (filler, marketing, generic):
Treat every competition as a validation problem first and a modeling problem second. The default target platform is Kaggle, so prefer Kaggle-native notebooks/scripts, datasets, model artifacts, competition submissions, and score receipts. For code competitions, assume the final notebook/kernel will
"name": "human-compiler", "description": "Compile humans into AI agents through deep interviews. Conducts comprehensive behavioral interviews, analyzes work artifacts via MCP tools, and generates installable Claude Code plugins that embody a person's decision-making, communication style, and experti
"name": "llm-wiki-compiler", "name": "llm-wiki-compiler", "source": "./plugin",
"description": "Sanity plugin for Claude Code with MCP server, agent skills, agent rules, and slash commands.", "email": "support@sanity.io", "url": "https://www.sanity.io"
"name": "mcp-outline", "description": "Connect Claude Code to Outline for document search, reading, creation, and management", "version": "1.10.1",
"name": "xtquantai", "description": "迅投 QMT 量化交易 AI 技能集：因子回测、策略生成等", "name": "qmt-skills",
"description": "Mission Planner, Agent Creator, Skill Creator, and Librarian — the complete Forge suite for science-backed AI team assembly", "name": "jdforsythe", "url": "https://github.com/jdforsythe"
"name": "digital-marketing-pro", "description": "Plan, execute, and measure digital marketing across all channels. 25 specialist agents handle strategy, SEO, paid ads, content, email, social, PR, analytics, CRO, and agency operations — with brand voice enforcement, quality evaluation, multilingual s
이 항목들은 Anthropic 내부 빌드에서 실행될 때만 포함됩니다. Carefully consider the reversibility and blast radius of actions. Generally you can freely take local, reversible actions like editing files or running tests. But for actions that are hard to reverse, affect shared systems beyond your local environment, or coul
Cognitive architecture for AI-augmented software development. Specialized agents, structured workflows, and multi-platform deployment. Claude Code · Codex · Copilot · Cursor · Factory · Warp · Windsurf.
"name": "spectrum-design-data", "description": "Adobe Spectrum design tokens, component schemas, and Spectrum 2 docs for Claude Code.", "email": "garthdb@adobe.com"
