Prompts
Prompt engineering and templates
skills-freshness
by steven-pribilinskiy
Scan skills for stale references: deprecated model IDs, outdated API endpoints, retired library versions. Run periodically or when you notice a skill using old tech. Also runs npx skills check for remote-source updates.
portrait-gallery
by i-kirito
Install, configure, run, and operate the Portrait Gallery AI outfit portrait system. Use when the user wants an AI agent to set up the gallery from GitHub, start or debug the local service, configure image/LLM API keys, generate daily outfit schedules, create custom portrait images, inspect the web gallery, or call the REST API for photo management.
caveman-mode
by team-horeca1
Personality override that makes Claude communicate like a caveman — terse, blunt, no filler. Use this skill for ALL responses when installed. This skill is always active. It applies to every single message Claude sends. Do not skip it. Do not "switch back" to normal mode. If this skill is loaded, caveman mode is ON.
浣跨敤IMYAI
by vegetpig
"浣跨敤IMYAI. Let Codex App call IMYAI official chat and image models by human-readable model name while Codex App remains the orchestrator and tool executor. Use when the user says 浣跨敤IMYAI, asks Codex App to use an IMYAI model for later replies, names an IMYAI model such as Claude Opus/Sonnet, Qwen, Gemini, Ava, GPT Image, Nano Banana, or asks for IMYAI image generation, login, JWT verification, model discovery, session mode, or IMYAI-first repair."
llm-architect
by ramishaheen
"Comprehensive guide for building unique LLM architectures and showcasing them on GitHub. Use for: building LLMs from scratch, fine-tuning open-source models, implementing RAG systems, setting up development environments, and creating portfolio projects."
agent-skills
by Xalior
🤖 AI Agent Skills Collection - Like vercel-labs/agent-skills, but cooler. Because emojis. 😎
claude-cli
by xjthy001
"Teaches agents how to use the Claude Code CLI (the claude command). Provides complete reference for invoking claude commands, automating with claude -p, configuring hooks/MCP/subagents/skills/settings/permissions, managing sessions (-c, -r), and working with CLAUDE.md memory. Use when: (1) an agent needs to call or invoke claude CLI, (2) scripting or automating with claude -p, (3) configuring settings, permissions, hooks, or MCP servers, (4) creating subagents or skills, (5) managing sessions and context, (6) any question about Claude Code CLI usage, flags, or features."
managing-claude-code-meta
by mikekelly
"MUST be loaded when setting up, installing, migrating, reviewing, auditing, or checking CLAUDE.md files in projects. Covers installing the promode CLAUDE.md into new projects, migrating existing CLAUDE.md content to READMEs (progressive disclosure), and auditing projects for conformance. Invoke PROACTIVELY when user mentions CLAUDE.md, project setup, agent configuration, or code meta files."
app-builder
by ncdevshiv
"Main application building orchestrator. Creates full-stack applications from natural language requests. Determines project type, selects tech stack, coordinates agents."
create-command
by blueif16
创建新的slash command，如果需要会自动初始化插件结构
create-hook
by blueif16
创建或添加event hooks到插件
prompt-optimizer
by gn00678465
Analyze and optimize user prompts through systematic deconstruction and framework-based reconstruction. Use when users request prompt improvement, optimization, refactoring, or restructuring. Triggers include "optimize this prompt", "improve this instruction", "refactor this prompt", "make this clearer", "restructure this prompt", or when analyzing poorly-structured prompts. This skill analyzes prompts for ambiguity, applies proven frameworks (B.R.O.K.E, T.R.A.C.E, C.L.E.A.R, etc.), and outputs high-precision, unambiguous prompts.
watchdog
by VictorVVedtion
"AI 会话免疫系统 - 环境健康感知，轻触提醒"
une-deux-post
by t-cmykh
Utiliser ce skill dès que Thomas demande de préparer, créer ou conformer un post pour une·deux (@une.deux), son média foot Instagram "le foot en deux touches". Couvre la production complète d'un post carrousel : script slides, fichier script.json prêt pour l'éditeur HTML, prompts photo style presse, et légende Instagram. Se déclenche sur "post pour une·deux", "on fait un post", "prépare un post sur [pays/joueur]", ou toute demande de contenu éditorial une·deux. Inclut la vérification factuelle obligatoire avant rédaction.
booking-com
by CodeDraig
Search and book hotels, apartments, and accommodations worldwide via Booking.com
ai-stock-board
by xiaomin4576-ui
美股 AI 科技股早报。以二级市场交易员/buy-side 视角,长期(6-12月)跟踪「美股 AI 产业链上下游核心十票 + QQQ」(英伟达、博通、台积电、超微、美光、迈威尔、Arista、Credo、相干、Vertiv + QQQ),每支给两个核心指标:①买入建议(买入/观望/回避+建议买入价)②6-12月目标价+预期收益。内置"预测台账→自动复盘→校准"闭环,让买入价/目标价越用越准。引擎是 Claude 本人(不用 DeepSeek),数据用 yfinance 真实行情。当用户说"跑美股AI早报/AI股票早报/科技股早报/英伟达们今天怎么看/ai-stock-board"或需要定时盯这组票时使用(本地出看板)。另:说"刷新云端美股看板/把美股早报推到线上/跑美股AI早报云端版/更新线上美股看板"时,走【云端手动刷新】——Claude 亲自研判一期并 push 到云端仓库,触发线上 Pages 重建+飞书推送。
save-conversation
by HartreeWorks
This skill should be used when the user asks to "save this conversation", "save conversation", "export conversation", "save our chat", "create a transcript", or mentions wanting to share or archive the current conversation. Exports Claude Code conversations to readable markdown files.
agentic-development
by campbellmcgregor
Conversational guidance for building software with AI agents, covering workflows, tool selection, prompt strategies, parallel agent management, and best practices based on real-world high-volume agentic development experience. Use this skill when users ask about setting up agentic workflows, choosing models, optimizing prompts, managing parallel agents, or improving agent output quality.
vibeads
by pooriaarab
"Contextual dev tool discovery for Claude Code -- surfaces relevant a16z portfolio tools based on what you're building"
structured-delivery
by janjaszczak
Provide structured outputs (plan/report/checklist/table) optimized for readability and reuse. Use when user requests a plan, report, checklist, template, or when high-risk requires explicit verification steps.
learn-plan
by xinzengxx
生成长期学习计划文件 learn-plan.md，并以三轮递进方式衔接 /learn-today 和 /learn-test
qavr-status
by kimasplund
Display QAVR (Q-Value Augmented Vector Retrieval) system status including mode, interaction counts, top memories by Q-value, and configuration.
notebooklm-connector
by Jiang-Yude
公開版 NotebookLM 連接器技能包。讓 Claude Code 與 Codex CLI 透過 MCP 直接對自己的 Google NotebookLM 內容做問答查詢，並可選用 ExportKit 把 NotebookLM 內容匯出回本地、用 nlm CLI 管理 sources。當使用者說「連 NotebookLM」「在 Claude Code 用 NotebookLM」「在 Codex 問 NotebookLM」「裝 notebooklm-connector」「設定 NotebookLM MCP」時觸發。第一次使用先跑安裝腳本，再依 references/mcp-setup.md 把 MCP 設定加到 Claude Code 或 Codex 的設定檔。授權：MIT。
prompt-engineer
by enoch-robinson
Prompt 工程最佳实践指南。当用户需要优化 AI 提示词、设计系统提示、提升 LLM 输出质量、或构建 AI 应用的提示策略时使用此技能。