Find awesome Agent Skills

code-simplifier

Simplifies and refines code for clarity, consistency, and maintainability while preserving all functionality. Focuses on recently modified code unless instructed otherwise.

plan-mode

完整复刻 Claude Code Plan Mode 架构。增强版：强制代码落地 + 显式暂停询问机制。

java-dev

Comprehensive Java development skill based on Alibaba Java Coding Guidelines (Songshan Edition). Use when writing, reviewing, or refactoring Java code to ensure compliance with industry best practices. Triggers on: (1) Writing new Java code (.java files), (2) Reviewing existing Java code, (3) Refactoring Java projects, (4) Database design with MySQL, (5) API design and implementation, (6) Unit testing, (7) Concurrent programming, (8) Security implementation, or any Java development tasks requiring adherence to coding standards.

react-dev

MUST BE USED PROACTIVELY whenever writing React or TypeScript code. This agent provides authoritative guidance on architecture, component design, state management, and performance optimization, ensuring production-level scalability and maintainability. [Next.js stack specific]

vue-dev

MUST BE USED PROACTIVELY whenever writing Vue 3 or TypeScript code. This agent provides authoritative guidance on Vue architecture, component design, state management, and performance optimization, ensuring production-level scalability and maintainability. [Vue 3 stack specific]

x-algo-pipeline

Explain the complete X recommendation algorithm pipeline. Use when users ask how posts are ranked, how the algorithm works, or want an overview of the recommendation system.

x-algo-engagement

Reference for X algorithm engagement types and signals. Use when analyzing engagement metrics, action predictions, or understanding what signals the algorithm tracks.

x-algo-ml

Explain the Phoenix ML model architecture for X recommendations. Use when users ask about embeddings, transformers, how predictions work, or ML model details.

x-algo-filters

Explain why posts get filtered from the X feed. Use when analyzing why a post was removed, not shown, or filtered out of recommendations.

x-algo-scoring

Calculate and explain X algorithm engagement scores. Use when analyzing post ranking, understanding score weights, engagement potential, or why one post ranks higher than another.

jetbrains-skill

Use the JetBrains IDE MCP Server (IntelliJ IDEA 2025.2+) for IDE-backed indexing, inspections, symbol info, project file operations, and structural refactoring. Use it when you need exact project-aware analysis or safe IDE refactors, rather than broad semantic repository discovery.

desktop-commander

Use Desktop Commander MCP (typically tools like `mcp__desktop-commander__*`) for local files, precise edits, search, and long-running processes on the machine. Use it when the task requires direct local execution, rather than IDE-only actions or remote platform operations.

shrimp-task-manager

Use Shrimp Task Manager MCP for structured task analysis, decomposition, execution guidance, tracking, and verification. Use it when the main need is to manage a task system with dependencies and completion checks, rather than just expose a reasoning chain.

vercel-skill

Use Vercel MCP for project discovery, deployments, build and runtime logs, toolbar threads, protected preview access, and Vercel documentation. Use it when the task centers on Vercel-hosted application state, rather than local-only execution.

ace-tool

Use Ace Tool MCP (`mcp__ace_tool__search_context`) for semantic repository discovery. Use it when the exact file, symbol, or implementation path is unknown, rather than starting with exact IDE search or direct file edits.

chrome-devtools

Use Chrome DevTools MCP for real-browser debugging: console and network inspection, runtime state checks, Lighthouse audits, and performance tracing. Use it when diagnosis and browser internals matter more than scripted cross-browser automation.

context7

Use Context7 MCP for up-to-date library and framework documentation tied to an exact library ID. Use it for versioned API usage, setup, migrations, and code examples, rather than generic documentation-site search.

github-skill

Use GitHub MCP for remote repository state: files, branches, commits, issues, pull requests, reviews, releases, and code search on GitHub. Use it when the task is centered on GitHub-hosted data, rather than a local checkout or IDE session.

frontend-dev-plain-web

Use when building a webpage with plain HTML and CSS, and only add JavaScript when the requirements clearly need interaction. It fits lightweight code-first work that needs doc lookup, real-image sourcing, browser verification, and required DevTools checks without framework-heavy processes.

sequential-thinking

Use Sequential Thinking MCP for explicit reasoning steps with revisions, branching, and uncertainty handling. Use it when the main need is to structure analysis, planning, or decision-making, rather than manage a persistent task system.

ref-skill

Use Ref MCP to search documentation sources and then read the exact returned URL. Use it for documentation-site discovery and page-level reading when the right doc page is not yet known, rather than versioned library-ID lookup.

ppt-workflow-builder

Use when creating a PPT or slide deck from a topic, article, research corpus, or rough brief, especially when you need a repeatable workflow for audience framing, JSON outline generation, slide research, planning drafts, and SVG-first slide production before assembling the final deck.

replicate-skill

Use Replicate MCP for models, predictions, trainings, deployments, and file management on Replicate. Use it when you need hosted AI workflows on Replicate, rather than general web research or local model execution.

playwright-skill

Use Playwright MCP for deterministic browser automation: navigate pages, fill forms, reproduce user flows, and run lightweight end-to-end interactions. Use it when scripted cross-browser execution matters more than DevTools-level debugging.

figma-skill

Use Figma MCP for design-to-code, code-to-Figma, design system search, Code Connect, and direct canvas edits. Use it when a task needs real Figma file context or writes to a Figma document, rather than generic browser automation.

grok-search

Use Grok Search MCP for planned web research: define intent, assess complexity, break the task into sub-queries, and then search, fetch, or map sources. Use it for multi-source research that benefits from explicit planning, and state which Grok model is being used whenever that is known.

deepwiki

Use DeepWiki MCP for repository-focused documentation from deepwiki.com. Use it when you need a fast overview of an external repository, architecture, or repo-specific concepts, rather than live library docs or broad web research.