ツール

Lets agents read and write Figma: extract variables, components and styles, create and edit screens, FigJam boards and slides on the canvas, and run accessibility checks and design-system audits. Needs a Figma personal access token, the Figma desktop app, Node.js and the companion plugin.

Simon Willison's website screenshot CLI: batch screenshots from YAML, record demo videos, export PDF and HTML, run JavaScript in pages to extract data, and dump accessibility trees. Built on Playwright; install a browser on first use.

Perplexity's official MCP with four tools: web search, grounded Q&A, deep research and reasoning, available remotely or run locally. Requires a Perplexity API key.

Gives agents on-demand guidance for the modern web platform, covering View Transitions, container queries, Popover, INP optimization, accessibility and more, with browser compatibility data so they avoid outdated code. Maintained by the Google Chrome team; currently in preview.

Gives agents a live local Excalidraw canvas: add, edit and delete elements to draw architecture diagrams and flowcharts, take screenshots to check the result, convert Mermaid and export .excalidraw files. Runs locally with no API key; also ships a CLI and an agent skill.

Runs open image models like FLUX, Qwen Image and Z-Image locally on a Mac: text-to-image, reference-based edits, inpainting and upscaling, all on-device. Needs Apple Silicon and uv; the first run downloads models and is slow. Check each model's license before commercial use.

Four tools for real-time web search, page content extraction, site mapping and full-site crawling, available as a remote server or run locally. Maintained by Tavily; needs an API key or OAuth sign-in, and a free account is available.

Supabase's official CLI: runs the full backend stack locally (Postgres, Auth, Storage, Edge Functions), manages database migrations, generates types and deploys functions. The local stack requires Docker.

Terraform and OpenTofu best practices from Anton Babenko: diagnoses the failure mode first, then guides testing, module structure, state management, CI/CD and security scanning. Works for AWS, Azure and GCP.

Uses hooks to enforce test-driven development in Claude Code: blocks implementation without a failing test, prevents code beyond what tests require, and can use lint results to push refactoring. Needs Node.js 22+. Still maintained, but the author suggests its successor Probity for new projects.

Reviews mobile and desktop app designs against Apple's Human Interface Guidelines, bundled as 123 pages from developer.apple.com, citing the exact guideline through five review lenses, and flags generic, templated choices. Works on screenshots, mockups or code for SwiftUI, Flutter or Electron.

WordPress's official skill collection: block development, block themes, plugin architecture, REST API, Interactivity API, WP-CLI, performance profiling and Playground, so agents write code to modern WordPress standards and avoid common security pitfalls.

Draws a code change or part of a codebase as an animated architecture or data-flow diagram. The agent writes a JSON graph, validates and renders it to SVG with the PR Lens CLI, then pushes it to a prlens.dev canvas or attaches it to a GitHub, GitLab or Bitbucket pull request.

Lets Claude Code and other agents call Antigravity CLI or Gemini CLI, using Gemini's large context window to analyze big files and whole codebases; with Gemini CLI it can also test code in a sandbox. Install and sign in to the CLI first. Not an official Google project.

Explains a codebase, feature, command or pull request to someone new to it as a PR Lens canvas: one plain sentence first, then a walkthrough that grows the diagram from two boxes to the whole picture, delivered as a shareable link. Requires the PR Lens skill and its npx CLI.

Stripe's official CLI: forward and trigger webhook events locally, tail API request logs in real time, and create, read, update or delete Stripe objects to debug payment integrations. Needs a Stripe account; run stripe login first.

MiniMax's official CLI, built for agents: chat with MiniMax models from the terminal and generate images, video and speech, plus speech transcription, image understanding and web search. Works with the international platform. Requires a MiniMax Token Plan subscription.

A native Go MCP server for Kubernetes and OpenShift: talks to the API server directly to create, read, update and delete any resource, view pod logs, exec commands and install Helm charts, without needing kubectl. Supports multiple clusters.

TypeUI's design fundamentals skill: general UI and interaction principles covering visual hierarchy, spacing, typography, 30 laws of interaction and WCAG accessibility. Needs Node.js to install; the TypeUI design-system library needs its hosted MCP and sign-in.

Translation management for Ruby/Rails: statically analyzes code to find missing and unused keys, normalizes locale files, fills gaps via Google Translate, DeepL or OpenAI, and exports CSV for translators. Requires the i18n gem (Rails default) and Bundler; machine translation needs an API key.

Turns Figma designs into production code in your project, using the design screenshot as the visual target and reusing existing components, design tokens and Code Connect mappings first. Maintained by Figma; needs the Figma MCP server (free plans get only 6 calls a month).

Figma's official skill collection: implement designs as code, map components with Code Connect, generate design-system libraries from code, and build screens and FigJam diagrams on the canvas. Needs the Figma MCP server; canvas writes are in beta and currently free.

Figma's official plugin: connects the remote Figma MCP server and bundles skills for design-to-code, Code Connect, and generating designs and libraries, letting agents read design context and write to the canvas. Needs a Figma account; Starter plans and View/Collab seats get 6 calls a month.

Figma's official skill that teaches agents to run Plugin API scripts via use_figma, creating and editing nodes, components, variables and auto layout directly on the canvas. Load it before any use_figma call. Needs a Figma account and the remote MCP server.