Explorar skills

'Generate concise, targeted pitch drafts for products, projects, or ideas tailored to a target audience. Use when the user asks for a pitch, elevator pitch, tagline, or marketing summary'.

Generate and navigate CODEMAP.md index files for large codebases. Creates hierarchical per-directory CODEMAP.md files containing simplified directory structure, file/subdirectory function summaries, domain annotations, task-to-file routing, file-level dependency tables, and directory-level key exports with source location annotations. Use this skill whenever the user wants to index a codebase for

Generate, review, and validate Apple App Store metadata (app name, subtitle, keywords, description, promotional text, release notes) against Apple's character limits and 2025-2026 ASO best practices. Trigger when the user mentions App Store metadata, ASO, App Store Connect, keyword research for iOS, screenshot captions, or release notes for an iOS / iPadOS / macOS / visionOS / tvOS / watchOS app.

拓（Tà）：通过深度引导式问答和本地语料解析，拓印普通人的"表达DNA"和"认知操作系统"，生成高保真的数字分身Skill。 核心特色：动态评估表达能力、渐进式提问漏斗、区分"想"与"说"、矛盾检测校准、强制完成度检查、渐进式验证循环。 触发词：「蒸馏我自己」「做一个XX的数字分身」「拓印XX的人格」「拓」。

Configure Claude Code to launch third-party LLM backends (GLM / DeepSeek / Qwen / Kimi / etc.) via their Anthropic-compatible endpoints. Creates per-provider launcher scripts (`~/.claude/claude-<name>.sh` and `.ps1`) plus matching shell aliases so the user can run `glm`, `ds`, `qwen`, etc. to start Claude Code with that backend. Trigger on: "add <provider> to claude code", "配一个 <模型> 启动器", "加一个 <模型

Build websites in the Swiss / International Typographic Style — Müller-Brockmann grids, disciplined typography, ink-on-paper restraint, and the modern agency-grade idioms (arrow-in-circle CTAs, eyebrow tags, type-as-architecture, ambient parallax, Lenis smooth scroll, custom cursor, oklch color, dark mode, view transitions) that make a Swiss-style site memorable without being noisy. Use when the b

Helps students identify professors for research opportunities by extracting publicly available faculty information including emails, profile links, and recent research papers.

Comprehensive security auditor for Claude Skills and MCP servers. Analyzes code for malicious patterns, suspicious behaviors, and security vulnerabilities. Provides detailed risk assessment and recommendations. Use when: evaluating new skills, auditing MCP servers, checking downloaded code, or verifying skill safety. Activate on: "security audit", "is this skill safe", "check this skill", "audit M

Inject a vanilla-JS annotation overlay into any HTML report or live web preview. Select text or click-drag a region → bubble for comments → yellow highlight saved to LocalStorage. One-click "Copy as Prompt" exports a structured markdown bundle to clipboard for the user to paste back to Claude Code. Tracks status per annotation (active / exported / done) so successive iteration rounds don't duplica

Generates a decision tree for project tasks. Forks every meaningful choice, simulates each branch forward up to 10 steps, scores outcomes against weighted objectives, and emits `tasks/forkcast.md` — a human-editable markdown decision tree with embedded mermaid diagram. Closes the gap LeCun's critique points at: LLMs lack world models, so we make the model's lookahead, confidence, and grounding *le

Scan a codebase to identify hand-rolled implementations that should be replaced by third-party libraries, identify missing capabilities, and detect code organization issues (directory structure, naming, circular deps, barrel bloat). Produce structured migration plans with Context7-verified recommendations. Use when analyzing technical debt, auditing dependency health, reviewing hand-rolled code, p

❄️高冷女王御姐 Skill，气场拉满杀伐果断，强势清醒的 AI 决策导师｜Claude Code Persona