fx skills, MCP servers, and subagents

August 31, 2026 · View on GitHub

Install checks Last checked Last commit

fx skills, MCP servers, and subagents

Everything you can plug into fx, the terminal coding agent from Vercel Labs. Every skill here installs with one /skills add line, checked against fx's own resolution rules every week.

fx reads SKILL.md folders from twelve directories, including .claude/skills and .agents/skills, so most of the agent-skill ecosystem already works in it unmodified. It also speaks MCP with a full OAuth client, runs subagents as child sessions, and embeds as a WebAssembly module. This list is what that buys you, sorted by the job you are trying to do.

⭐ Featured skill
🚀 Where do I start?
fx skills
fx MCP servers
Gateways and bridges
Ports and packaging
Embedding fx
Good to know
Full catalog: every verified fx skill (1259) in CATALOG.md

⭐ Featured skill

Search YouTube and pull transcripts with youtube-full by ZeroPointRepo. Transcripts, video and channel search, playlists, and within-channel search, in one skill. No Google API key, and nothing to install or maintain. 586★.

/skills add ZeroPointRepo/youtube-skills --skill youtube-full

🚀 Where do I start?

1. Install fx. One binary, about 8 MB, no runtime.

curl -fsSL https://fx.sh/setup.sh | bash

2. Add a skill. Every entry below follows the same shape, typed inside an fx session:

/skills add <owner>/<repo> --skill <name>

fx clones the repo, finds the matching SKILL.md, and copies just that folder into ~/.fx/skills. Drop --skill to take every skill in the repo. npx skills add owner/repo typed as a shell command works too, and fx installs it itself without running npx.

3. Pick the one that fixes today. Not sure? Start with systematic-debugging if something is broken, or frontend-design if you are staring at a blank page.

fx is marked Experimental by its own authors and moves fast, five releases in its first five days. Re-check a workflow after an upgrade.

fx skills

Design and build interfaces

Give a new UI an actual point of view with frontend-design by anthropics. Aesthetic direction, typography, and colour choices that do not read as template output. Anthropic's own. 172.7k★.
Install
```
/skills add anthropics/skills --skill frontend-design
```
Have your UI reviewed against the Web Interface Guidelines with web-design-guidelines by vercel-labs. Accessibility, focus states, motion, and layout, checked as a pass over the code you just wrote. 30.7k★.
Install
```
/skills add vercel-labs/agent-skills --skill web-design-guidelines
```
Ship a landing page that does not look generated with design-taste-frontend by Leonxlnx. Reads the brief, picks a direction, and holds a line on spacing, type scale, and restraint. 82.7k★, MIT.
Install
```
/skills add Leonxlnx/taste-skill --skill design-taste-frontend
```
Get the small details that make software feel finished with emil-design-eng by emilkowalski. Emil Kowalski on component polish, animation decisions, and the invisible parts. 34.0k★, MIT.
Install
```
/skills add emilkowalski/skills --skill emil-design-eng
```
Build gesture and spring motion that feels native with apple-design by emilkowalski. Apple's interface and motion approach translated for the web: drags, swipes, sheets, springs. 34.0k★, MIT.
Install
```
/skills add emilkowalski/skills --skill apple-design
```
Design dashboards and admin panels properly with interface-design by Dammyjay93. Craft-first patterns for data tables, settings pages, and dense product surfaces. 5.6k★, MIT.
Install
```
/skills add Dammyjay93/interface-design --skill interface-design
```
Fix the accessibility bugs already in your markup with fixing-accessibility by ibelick. ARIA labels, keyboard traps, focus management, contrast, and form error wiring. 7.9k★, MIT.
Install
```
/skills add ibelick/ui-skills --skill fixing-accessibility
```
Build custom widgets that keyboards can drive with better-accessibility by jakubkrehel. Accessibility engineering for components, not a lint pass over finished HTML. 4.7k★, MIT.
Install
```
/skills add jakubkrehel/skills --skill better-accessibility
```
Pull a design system out of a website you like with extract-design-system by arvindrk. Extracts primitives from a public URL and writes starter token files for your project. 194★, MIT.
Install
```
/skills add arvindrk/extract-design-system --skill extract-design-system
```
Turn a Stitch project into a DESIGN.md the agent reads with design-md by google-labs-code. Synthesises a semantic design system file, so later prompts inherit the rules instead of re-arguing them. 8.2k★, Apache-2.0.
Install
```
/skills add google-labs-code/stitch-skills --skill design-md
```
Build accessible, responsive UI to a production bar with frontend-ui-engineering by addyosmani. WCAG targets, responsive behaviour, and component structure, from Addy Osmani. 91.0k★, MIT.
Install
```
/skills add addyosmani/agent-skills --skill frontend-ui-engineering
```
Build with HeroUI v3 components with heroui-react by heroui-inc. Buttons, modals, forms, and cards on Tailwind v4 and React Aria, installed and wired correctly. 30.5k★, Apache-2.0.
Install
```
/skills add heroui-inc/heroui --skill heroui-react
```

Review and refactor code

Review everything since a commit, branch, or merge-base with code-review by mattpocock. Two axes at once: does it follow the repo's documented standards, and does it match the spec. 242.4k★, MIT.
Install
```
/skills add mattpocock/skills --skill code-review
```
Get a shared vocabulary for module design with codebase-design by mattpocock. Where a seam goes, how deep a module should be, and how to name the thing you just built. 242.4k★, MIT.
Install
```
/skills add mattpocock/skills --skill codebase-design
```
Have the agent review its own work before you see it with requesting-code-review by obra. Runs before a merge or a hand-off and checks the work against what was actually asked for. 279.9k★, MIT.
Install
```
/skills add obra/superpowers --skill requesting-code-review
```
Push back on review feedback that is wrong with receiving-code-review by obra. Verifies a suggestion technically before implementing it, instead of complying with everything. 279.9k★, MIT.
Install
```
/skills add obra/superpowers --skill receiving-code-review
```
Run a multi-axis review before merging with code-review-and-quality by addyosmani. Correctness, readability, tests, and risk, scored separately rather than as one vibe. 91.0k★, MIT.
Install
```
/skills add addyosmani/agent-skills --skill code-review-and-quality
```
Get an unreasonably strict maintainability review with thermo-nuclear-code-quality-review by cursor. Hunts giant files, leaky abstractions, and spaghetti conditionals. Use it when you want it to hurt. 6.3k★.
Install
```
/skills add cursor/plugins --skill thermo-nuclear-code-quality-review
```
Run CodeRabbit review from inside the session with code-review by coderabbitai. Fires on an explicit request, and on its own when the agent thinks a review is due. 162★, MIT.
Install
```
/skills add coderabbitai/skills --skill code-review
```
Review staged changes or the whole working tree with code-reviewer by google-gemini. Google's review skill, works on local diffs as well as a PR. 106.8k★, Apache-2.0.
Install
```
/skills add google-gemini/gemini-cli --skill code-reviewer
```
Review the current git diff with a senior lens with code-review-expert by sanyuan0704. SOLID violations, security risk, and concrete rewrites rather than adjectives. 3.9k★, MIT.
Install
```
/skills add sanyuan0704/sanyuan-skills --skill code-review-expert
```
Refactor without changing behaviour with refactor by github. Extract functions, break up god objects, rename honestly, keep the tests green. 38.5k★, MIT.
Install
```
/skills add github/awesome-copilot --skill refactor
```
Bracket a structural change with verification with safe-refactor by JuliusBrussee. Proves behaviour before and after the move, so a "pure refactor" is actually pure. 102.0k★.
Install
```
/skills add JuliusBrussee/caveman --skill safe-refactor
```

Test and debug

Stop guessing at bugs with systematic-debugging by obra. Forces a hypothesis and a reproduction before any fix is proposed. The most-installed debugging skill there is. 279.9k★, MIT.
Install
```
/skills add obra/superpowers --skill systematic-debugging
```
Get to root cause when the build breaks with debugging-and-error-recovery by addyosmani. A repeatable path from failing test to the actual cause, instead of shotgun edits. 91.0k★, MIT.
Install
```
/skills add addyosmani/agent-skills --skill debugging-and-error-recovery
```
Chase competing hypotheses at the same time with parallel-debugging by wshobson. Splits the investigation, collects evidence per branch, then arbitrates on the root cause. 39.3k★, MIT.
Install
```
/skills add wshobson/agents --skill parallel-debugging
```
Drive your local app with Playwright with webapp-testing by anthropics. Click through the running app, screenshot it, and verify the frontend actually does what you think. 172.7k★.
Install
```
/skills add anthropics/skills --skill webapp-testing
```
Inspect a real browser over Chrome DevTools MCP with browser-testing-with-devtools by addyosmani. DOM, network, console, and performance traces from the page as it actually renders. 91.0k★, MIT.
Install
```
/skills add addyosmani/agent-skills --skill browser-testing-with-devtools
```
Turn a described scenario into a Playwright test with playwright-generate-test by github. Writes the spec from a plain-language walkthrough using Playwright MCP. 38.5k★, MIT.
Install
```
/skills add github/awesome-copilot --skill playwright-generate-test
```
Build an end-to-end suite that is not flaky with e2e-testing-patterns by wshobson. Playwright and Cypress patterns for suites you will still trust in three months. 39.3k★, MIT.
Install
```
/skills add wshobson/agents --skill e2e-testing-patterns
```
Write pytest suites that survive refactors with python-testing-patterns by wshobson. Fixtures, mocking boundaries, and a TDD loop that fits real projects. 39.3k★, MIT.
Install
```
/skills add wshobson/agents --skill python-testing-patterns
```
Test JavaScript with Jest, Vitest, and Testing Library with javascript-testing-patterns by wshobson. Unit, integration, and mocking strategy without testing implementation details. 39.3k★, MIT.
Install
```
/skills add wshobson/agents --skill javascript-testing-patterns
```
Verify with VoiceOver, NVDA, and JAWS with screen-reader-testing by wshobson. The part of accessibility work an automated audit cannot do for you. 39.3k★, MIT.
Install
```
/skills add wshobson/agents --skill screen-reader-testing
```

Databases and auth

Work with Supabase without guessing the API with supabase by supabase. Database, Auth, Edge Functions, Realtime, Storage, and the SSR integrations. Maintained by Supabase. 2.6k★, MIT.
Install
```
/skills add supabase/agent-skills --skill supabase
```
Get Postgres right before you write the migration with supabase-postgres-best-practices by supabase. Schema, index, and RLS practice for Postgres anywhere, not only on Supabase. 2.6k★, MIT.
Install
```
/skills add supabase/agent-skills --skill supabase-postgres-best-practices
```
Create and operate a Prisma Postgres database with prisma-postgres by prisma. Console, create-db CLI, and the Management API, from Prisma. 52★, MIT.
Install
```
/skills add prisma/skills --skill prisma-postgres
```
Write Prisma Client queries correctly with prisma-client-api by prisma. Model queries, filters, operators, and the client methods people usually look up twice. 52★, MIT.
Install
```
/skills add prisma/skills --skill prisma-client-api
```
Work with Neon Postgres and branching with neon-postgres by neondatabase. Neon's own guidance for serverless Postgres, branches, and connection handling. 84★, Apache-2.0.
Install
```
/skills add neondatabase/agent-skills --skill neon-postgres
```
Plan and review MySQL schema and indexes with mysql by planetscale. InnoDB-aware schema design, query tuning, and transaction behaviour from PlanetScale. 641★, MIT.
Install
```
/skills add planetscale/database-skills --skill mysql
```
Use the Postgres-only features you are ignoring with postgresql-optimization by github. JSONB, partial indexes, CTE behaviour, and the planner details that decide query cost. 38.5k★, MIT.
Install
```
/skills add github/awesome-copilot --skill postgresql-optimization
```
Migrate a live database with zero downtime with database-migration by wshobson. Expand-contract sequencing, backfills, and a rollback that actually works. 39.3k★, MIT.
Install
```
/skills add wshobson/agents --skill database-migration
```
Model, query, and secure Firestore with firebase-firestore by firebase. Data modelling, security rules, indexes, and the SDK wiring, from Firebase. 430★, Apache-2.0.
Install
```
/skills add firebase/agent-skills --skill firebase-firestore
```
Stop an agent deploying to prod by accident with convex-deploy-guard by get-convex. Classifies the target deployment and demands fresh consent before anything touches production. 51★.
Install
```
/skills add get-convex/agent-skills --skill convex-deploy-guard
```
Wire Clerk auth into Next.js properly with clerk-nextjs-patterns by clerk. Middleware, Server Actions, and caching behaviour, from Clerk. 69★.
Install
```
/skills add clerk/skills --skill clerk-nextjs-patterns
```
Harden a Better Auth setup with better-auth-security-best-practices by better-auth. Rate limits, CSRF, trusted origins, session and cookie flags, OAuth token encryption. 210★.
Install
```
/skills add better-auth/skills --skill better-auth-security-best-practices
```

Deploy and run in production

Deploy and get the preview link back with deploy-to-vercel by vercel-labs. "Push this live" as one instruction. Vercel's own deployment skill. 30.7k★.
Install
```
/skills add vercel-labs/agent-skills --skill deploy-to-vercel
```
Deploy from CI with a token instead of a login with vercel-cli-with-tokens by vercel-labs. Token-based Vercel CLI use for pipelines and headless sessions. 30.7k★.
Install
```
/skills add vercel-labs/agent-skills --skill vercel-cli-with-tokens
```
Cut what a deployed project costs to run with vercel-optimize by vercel-labs. Pulls real usage and config, then names the specific changes worth making. 30.7k★.
Install
```
/skills add vercel-labs/agent-skills --skill vercel-optimize
```
Execute an Azure deployment that is already planned with azure-deploy by microsoft. Runs the deployment for a project with an existing plan and infrastructure files. 1.4k★, MIT.
Install
```
/skills add microsoft/azure-skills --skill azure-deploy
```
Stand up a production-ready AKS cluster with azure-kubernetes by microsoft. Day-0 checklist, SKU choice, private API server, and the networking decisions up front. 1.4k★, MIT.
Install
```
/skills add microsoft/azure-skills --skill azure-kubernetes
```
Deploy a Next.js or Angular app with SSR with firebase-app-hosting-basics by firebase. Firebase App Hosting for full-stack apps, not just static output. 430★, Apache-2.0.
Install
```
/skills add firebase/agent-skills --skill firebase-app-hosting-basics
```
Build stateful agents on Cloudflare Workers with agents-sdk by cloudflare. Durable workflows, WebSocket apps, scheduled tasks, and MCP servers on the Agents SDK. 2.7k★, Apache-2.0.
Install
```
/skills add cloudflare/skills --skill agents-sdk
```
Write Terraform HashiCorp would accept with terraform-style-guide by hashicorp. Official HCL style conventions, applied while the code is being written. 855★, MPL-2.0.
Install
```
/skills add hashicorp/agent-skills --skill terraform-style-guide
```
Build reusable modules across AWS, Azure, and GCP with terraform-module-library by wshobson. Module boundaries, variables, and outputs that survive being used twice. 39.3k★, MIT.
Install
```
/skills add wshobson/agents --skill terraform-module-library
```
Get a small, correct Dockerfile with multi-stage-dockerfile by github. Multi-stage builds for any language, with the caching layers in the right order. 38.5k★, MIT.
Install
```
/skills add github/awesome-copilot --skill multi-stage-dockerfile
```
Write manifests, RBAC, and pod security with kubernetes-specialist by Jeffallan. Deployments, service accounts, and policies for real cluster work. 11.3k★, MIT.
Install
```
/skills add Jeffallan/claude-skills --skill kubernetes-specialist
```
Design a pipeline with real gates with deployment-pipeline-design by wshobson. Multi-stage CI/CD, approvals, security checks, and zero-downtime rollout. 39.3k★, MIT.
Install
```
/skills add wshobson/agents --skill deployment-pipeline-design
```

Browse the web and gather evidence

Give the agent a browser it can drive with agent-browser by vercel-labs. Navigate, fill forms, click, screenshot, and extract. Vercel's browser automation CLI for agents. 41.6k★, Apache-2.0.
Install
```
/skills add vercel-labs/agent-browser --skill agent-browser
```
Scrape and crawl sites from the session with firecrawl-agent by firecrawl. Firecrawl's own CLI skill for turning pages into clean structured text. 614★.
Install
```
/skills add firecrawl/cli --skill firecrawl-agent
```
Rebuild a site's design system from evidence with firecrawl-website-design-clone by firecrawl. Scrapes the real page, then writes colours, type, spacing, and components into a DESIGN.md. 145★, ISC.
Install
```
/skills add firecrawl/firecrawl-workflows --skill firecrawl-website-design-clone
```
Audit a site's SEO from a live crawl with firecrawl-seo-audit by firecrawl. Metadata, headings, site structure, and keyword gaps against the pages as served. 145★, ISC.
Install
```
/skills add firecrawl/firecrawl-workflows --skill firecrawl-seo-audit
```
Automate and profile Chrome over MCP with chrome-devtools by github. Interact with pages, capture screenshots, and read performance traces. 38.5k★, MIT.
Install
```
/skills add github/awesome-copilot --skill chrome-devtools
```
Use a real browser with a real fingerprint with browser-mcp-agent by antibrow. Launch, navigate, click, and extract with a persistent kernel-level device profile. 10★, MIT.
Install
```
/skills add antibrow/anti-detect-browser-skills --skill browser-mcp-agent
```
Search the web with citations attached with tavily-research by tavily-ai. Tavily's research skill, built for answers you can trace back. 468★, MIT.
Install
```
/skills add tavily-ai/skills --skill tavily-research
```
Run an exhaustive multi-source investigation with parallel-deep-research by parallel-web. Deliberately slow and thorough. Reach for it only when a quick search will not do. 72★, MIT.
Install
```
/skills add parallel-web/parallel-agent-skills --skill parallel-deep-research
```

Documents, diagrams, and slides

Read, split, merge, and fill PDFs with pdf by anthropics. Text and table extraction, page surgery, and form filling. Anthropic's own. 172.7k★.
Install
```
/skills add anthropics/skills --skill pdf
```
Work on PDFs where the layout matters with pdf by openai. Renders pages and checks them visually instead of trusting a text dump. OpenAI's own. 25.3k★.
Install
```
/skills add openai/skills --skill pdf
```
Pull tables and metadata out of a PDF with pdf-extraction by claude-office-skills. pdfplumber under the hood, for the documents that defeat a plain text extract. 427★, MIT.
Install
```
/skills add claude-office-skills/skills --skill pdf-extraction
```
Build and edit real spreadsheets with excel-automation by claude-office-skills. Formulas, formatting, and multi-sheet workbooks rather than a CSV pretending to be one. 427★, MIT.
Install
```
/skills add claude-office-skills/skills --skill excel-automation
```
Convert a PDF into Word, Markdown, or LaTeX with pdf-converter by tanis90. MinerU-powered, and it handles scanned pages through OCR. 54★.
Install
```
/skills add tanis90/pdf-converter-mineru --skill pdf-converter
```
Drive pdftk from the shell with pdftk-server by github. Merge, split, rotate, encrypt, and stamp, for when a library is overkill. 38.5k★, MIT.
Install
```
/skills add github/awesome-copilot --skill pdftk-server
```
Create a Google Slides deck and fill it with recipe-create-presentation by googleworkspace. Google's own Workspace CLI skill, one of dozens in the same repo. 30.7k★, Apache-2.0.
Install
```
/skills add googleworkspace/cli --skill recipe-create-presentation
```
Draw the architecture you just described with design-doc-mermaid by SpillwaveSolutions. Activity, sequence, deployment, and architecture diagrams in Mermaid, from text or from source. 159★.
Install
```
/skills add SpillwaveSolutions/design-doc-mermaid --skill design-doc-mermaid
```
Get an Excalidraw diagram from a description with excalidraw-diagram-generator by github. Hand-drawn-style flowcharts and system diagrams you can keep editing. 38.5k★, MIT.
Install
```
/skills add github/awesome-copilot --skill excalidraw-diagram-generator
```

Images, video, and audio

Turn a product URL into a launch video with product-launch-video by heygen-com. Feed it a marketing page, a script, or a brief and get a finished promo. 43.3k★, Apache-2.0.
Install
```
/skills add heygen-com/hyperframes --skill product-launch-video
```
Turn a pull request into an explainer video with pr-to-video by heygen-com. Point it at a PR URL and it narrates the change with the diff on screen. 43.3k★, Apache-2.0.
Install
```
/skills add heygen-com/hyperframes --skill pr-to-video
```
Turn a changelog into a branded clip with changelog-video by heygen-com. Square, about a minute, voiced, with mock-UI visualisations. 43.3k★, Apache-2.0.
Install
```
/skills add heygen-com/hyperframes --skill changelog-video
```
Generate video from a prompt with ai-video-generation by prime-skills. A router across RunComfy's video models, so you describe the shot rather than pick the model. 39★, MIT.
Install
```
/skills add prime-skills/runcomfy-agent-skills --skill ai-video-generation
```
Generate and edit images from the session with ai-image-generation by prime-skills. Same router idea for stills: generation, edits, and variants. 39★, MIT.
Install
```
/skills add prime-skills/runcomfy-agent-skills --skill ai-image-generation
```
Generate images across eleven providers with baoyu-image-gen by JimLiu. GPT Image 2, Google, OpenRouter, DashScope, GLM-Image, MiniMax, Seedream, Replicate, and more behind one skill. 25.5k★, MIT.
Install
```
/skills add JimLiu/baoyu-skills --skill baoyu-image-gen
```
Make a short video with no API key setup with videoagent-video-studio by pexoai. Text-to-video, image-to-video, and reference-based generation. 776★, MIT.
Install
```
/skills add pexoai/pexo-skills --skill videoagent-video-studio
```
Make a poster or a print-quality PNG with canvas-design by anthropics. Real design philosophy applied to static artwork, output as .png or .pdf. 172.7k★.
Install
```
/skills add anthropics/skills --skill canvas-design
```
Make a GIF that Slack will actually accept with slack-gif-creator by anthropics. Knows the size and dimension limits, and validates before you upload. 172.7k★.
Install
```
/skills add anthropics/skills --skill slack-gif-creator
```
Batch-process images with ImageMagick with image-manipulation-image-magick by github. Resize, convert, strip metadata, and run the whole folder. 38.5k★, MIT.
Install
```
/skills add github/awesome-copilot --skill image-manipulation-image-magick
```

Languages and frameworks

Write React and Next.js the way Vercel does with vercel-react-best-practices by vercel-labs. Performance guidance from Vercel Engineering, applied while the code is written. 30.7k★.
Install
```
/skills add vercel-labs/agent-skills --skill vercel-react-best-practices
```
Compose React components that scale with vercel-composition-patterns by vercel-labs. The refactor patterns for when prop drilling has quietly become the architecture. 30.7k★.
Install
```
/skills add vercel-labs/agent-skills --skill vercel-composition-patterns
```
Build React Native and Expo apps well with vercel-react-native-skills by vercel-labs. Vercel's mobile performance guidance for the same stack. 30.7k★.
Install
```
/skills add vercel-labs/agent-skills --skill vercel-react-native-skills
```
Master the App Router with nextjs-app-router-patterns by wshobson. Server Components, streaming, parallel routes, and the data fetching that goes with them. 39.3k★, MIT.
Install
```
/skills add wshobson/agents --skill nextjs-app-router-patterns
```
Use the type system on purpose with typescript-advanced-types by wshobson. Generics, conditional and mapped types, template literals, and when not to reach for them. 39.3k★, MIT.
Install
```
/skills add wshobson/agents --skill typescript-advanced-types
```
Set up Python with uv, ruff, and ty with modern-python by trailofbits. Trail of Bits' modern toolchain, including migrating off pip and Poetry. 6.9k★, CC-BY-SA-4.0.
Install
```
/skills add trailofbits/skills --skill modern-python
```
Find out why the Python is slow with python-performance-optimization by wshobson. cProfile and memory profiling first, optimisation second. 39.3k★, MIT.
Install
```
/skills add wshobson/agents --skill python-performance-optimization
```
Write Go tests that catch real failures with golang-testing by samber. Table-driven tests, testify suites, fuzzing, goroutine leak detection, coverage. 3.1k★, MIT.
Install
```
/skills add samber/cc-skills-golang --skill golang-testing
```
Write idiomatic Go with golang-design-patterns by samber. Functional options, error cascades, resource lifecycle, graceful shutdown. 3.1k★, MIT.
Install
```
/skills add samber/cc-skills-golang --skill golang-design-patterns
```
Get async Rust right with rust-async-patterns by wshobson. Tokio, async traits, error handling, and the concurrency patterns that compile. 39.3k★, MIT.
Install
```
/skills add wshobson/agents --skill rust-async-patterns
```
Write Rust to Apollo's handbook with rust-best-practices by apollographql. Idiomatic Rust from the team that maintains the Apollo Router. 108★, MIT.
Install
```
/skills add apollographql/skills --skill rust-best-practices
```
Write SwiftUI that does not re-render everything with swiftui-expert-skill by AvdLee. State flow with @Observable, view composition, invalidation, and list performance. 3.5k★, MIT.
Install
```
/skills add AvdLee/SwiftUI-Agent-Skill --skill swiftui-expert-skill
```
Review SwiftUI against modern APIs with swiftui-pro by twostraws. Paul Hudson's review skill for maintainability and performance. 4.6k★, MIT.
Install
```
/skills add twostraws/SwiftUI-Agent-Skill --skill swiftui-pro
```
Fix Swift concurrency errors and migrate to Swift 6 with swift-concurrency by AvdLee. Actors, @MainActor isolation, and moving callbacks to async/await. 1.6k★, MIT.
Install
```
/skills add AvdLee/Swift-Concurrency-Agent-Skill --skill swift-concurrency
```
Fix FPS, TTI, and bundle size in React Native with react-native-best-practices by callstackincubator. Callstack's performance guidance: re-renders, memory leaks, animations. 1.6k★, MIT.
Install
```
/skills add callstackincubator/agent-skills --skill react-native-best-practices
```
Test Vue apps with Vitest and Test Utils with vue-testing-best-practices by vuejs-ai. Component testing, mocking, and Playwright for the end-to-end layer. 2.8k★, MIT.
Install
```
/skills add vuejs-ai/skills --skill vue-testing-best-practices
```
Animate React with GSAP correctly with gsap-react by greensock. The official useGSAP hook, refs, contexts, and cleanup, from GreenSock. 14.6k★, MIT.
Install
```
/skills add greensock/gsap-skills --skill gsap-react
```

Security

Get a real vulnerability review of a change with security-review by getsentry. Sentry's own security review skill, aimed at the diff rather than the whole repo. 963★, Apache-2.0.
Install
```
/skills add getsentry/skills --skill security-review
```
Harden anything that takes untrusted input with security-and-hardening by addyosmani. Input handling, auth, storage, and third-party integrations. 91.0k★, MIT.
Install
```
/skills add addyosmani/agent-skills --skill security-and-hardening
```
Review security per language and framework with security-best-practices by openai. OpenAI's own review skill, deliberately opt-in rather than always-on. 25.3k★.
Install
```
/skills add openai/skills --skill security-best-practices
```
Audit Firestore and Storage rules with firebase-security-rules-auditor by firebase. Privilege escalation, role bypass, create-versus-update gaps, and resource exhaustion. 430★, Apache-2.0.
Install
```
/skills add firebase/agent-skills --skill firebase-security-rules-auditor
```
Secure an API you are about to ship with api-security-best-practices by sickn33. AuthN and authZ, validation, rate limiting, and the usual API-specific holes. 45.7k★, MIT.
Install
```
/skills add sickn33/agentic-awesome-skills --skill api-security-best-practices
```
Lock down a Kubernetes cluster with k8s-security-policies by wshobson. NetworkPolicy, pod security, and RBAC written for production rather than a demo. 39.3k★, MIT.
Install
```
/skills add wshobson/agents --skill k8s-security-policies
```
Write smart contracts that do not get drained with solidity-security by wshobson. The common vulnerability classes and the patterns that avoid them. 39.3k★, MIT.
Install
```
/skills add wshobson/agents --skill solidity-security
```
Get security guidance for a Google Cloud workload with google-cloud-waf-security by google. Google's own skill, keyed to the Well-Architected security pillar. 19.0k★, Apache-2.0.
Install
```
/skills add google/skills --skill google-cloud-waf-security
```
Vet a skill before you install it with skill-vetter by UseAI-pro. Reads a third-party SKILL.md for prompt injection and unsafe instructions first. 71★, MIT.
Install
```
/skills add UseAI-pro/openclaw-skills-security --skill skill-vetter
```

Plan work and manage git

Write the spec before any code exists with spec-driven-development by addyosmani. For a new project or a big feature where the requirements are still a conversation. 91.0k★, MIT.
Install
```
/skills add addyosmani/agent-skills --skill spec-driven-development
```
Break a spec into ordered, buildable tasks with planning-and-task-breakdown by addyosmani. For when the work is clear but too large to hold in one turn. 91.0k★, MIT.
Install
```
/skills add addyosmani/agent-skills --skill planning-and-task-breakdown
```
Keep PRODUCT.md and TECH.md true while you build with spec-driven-implementation by warpdotdev. Warp's spec-first loop, where the spec is updated as the implementation moves. 482★, MIT.
Install
```
/skills add warpdotdev/common-skills --skill spec-driven-implementation
```
Survive a context reset mid-task with pi-planning-with-files by OthmanAdi. Plans live in markdown on disk, so compaction or a crash does not lose the thread. 26.5k★, MIT.
Install
```
/skills add OthmanAdi/planning-with-files --skill pi-planning-with-files
```
Write a PRD with user stories and acceptance criteria with prd by github. Executive summary through to the edge cases, in a shape engineers can build from. 38.5k★, MIT.
Install
```
/skills add github/awesome-copilot --skill prd
```
Work on a feature without touching your workspace with using-git-worktrees by obra. Sets up an isolated worktree before a long change begins. 279.9k★, MIT.
Install
```
/skills add obra/superpowers --skill using-git-worktrees
```
Block the destructive git commands with git-guardrails-claude-code by mattpocock. Hooks that refuse push, reset --hard, clean, and branch -D before they run. 242.4k★, MIT.
Install
```
/skills add mattpocock/skills --skill git-guardrails-claude-code
```
Write conventional commit messages with conventional-commit by github. Structured analysis of the diff, then a message that matches it. 38.5k★, MIT.
Install
```
/skills add github/awesome-copilot --skill conventional-commit
```
Stage intelligently and commit with git-commit by github. Reads the working tree, groups related changes, and writes the message. 38.5k★, MIT.
Install
```
/skills add github/awesome-copilot --skill git-commit
```

Marketing, SEO, and writing

Find out why the site is not ranking with seo-audit by coreyhaines31. Technical and on-page SEO diagnosis rather than a keyword list. 46.2k★, MIT.
Install
```
/skills add coreyhaines31/marketingskills --skill seo-audit
```
Build template-driven pages at scale with programmatic-seo by coreyhaines31. Directory and comparison page programmes that do not read as doorway spam. 46.2k★, MIT.
Install
```
/skills add coreyhaines31/marketingskills --skill programmatic-seo
```
Get cited inside AI answers with ai-seo by coreyhaines31. AEO and GEO: structuring content so a model quotes you rather than your competitor. 46.2k★, MIT.
Install
```
/skills add coreyhaines31/marketingskills --skill ai-seo
```
Design an experiment worth running with ab-testing by coreyhaines31. Sample size, guardrail metrics, and a stopping rule agreed in advance. 46.2k★, MIT.
Install
```
/skills add coreyhaines31/marketingskills --skill ab-testing
```
Fix meta tags, structured data, and sitemaps with seo by addyosmani. The mechanical layer of search visibility, done properly. 2.7k★, MIT.
Install
```
/skills add addyosmani/web-quality-skills --skill seo
```
Optimise for search and for generative engines at once with seo-geo by ReScienceLab. One pass covering both classic SERP and AI answer surfaces. 1.7k★, Apache-2.0.
Install
```
/skills add ReScienceLab/opc-skills --skill seo-geo
```
Have your docs reviewed for voice and tone with writing-guidelines by vercel-labs. Vercel's writing guidelines applied as a review pass over prose. 30.7k★.
Install
```
/skills add vercel-labs/agent-skills --skill writing-guidelines
```
Write a SKILL.md or AGENTS.md that works with writing-for-agents by mattpocock. How to write documents whose reader is a model, not a person. 242.4k★, MIT.
Install
```
/skills add mattpocock/skills --skill writing-for-agents
```
Write the README, runbook, or onboarding guide with documentation by anthropics. Anthropic's documentation skill for the docs nobody volunteers to write. 23.8k★, Apache-2.0.
Install
```
/skills add anthropics/knowledge-work-plugins --skill documentation
```
Structure docs with Diataxis with documentation-writer by github. Tutorial, how-to, reference, and explanation kept as separate jobs. 38.5k★, MIT.
Install
```
/skills add github/awesome-copilot --skill documentation-writer
```

Connect the tools you already use

Drive Notion from the command line with notion-cli by makenotion. Notion's own ntn CLI skill for pages, databases, and workers. 161★, MIT.
Install
```
/skills add makenotion/skills --skill notion-cli
```
Call the Notion REST API directly with notion-api by intellectronica. For the operations the CLI does not cover. 290★, CC0-1.0.
Install
```
/skills add intellectronica/agent-skills --skill notion-api
```
Manage Linear issues from the terminal with linear-cli by schpet. Create, move, and comment on issues without leaving the session. 938★, ISC.
Install
```
/skills add schpet/linear-cli --skill linear-cli
```
Read and update the Linear issue you are working on with orca-linear by stablyai. Reads the linked issue for the current branch and keeps it current. 57.9k★, MIT.
Install
```
/skills add stablyai/orca --skill orca-linear
```
Wire Linear Release into CI with linear-release-setup by linear. Linear's own skill for generating the pipeline configuration. 62★, MIT.
Install
```
/skills add linear/linear-release --skill linear-release-setup
```
Create and triage GitHub issues over MCP with github-issues by github. Bug reports, feature requests, and bulk updates from the session. 38.5k★, MIT.
Install
```
/skills add github/awesome-copilot --skill github-issues
```
Turn a Gmail thread into a Google Task with gws-workflow-email-to-task by googleworkspace. One of a large set of Workspace recipes in Google's own CLI repo. 30.7k★, Apache-2.0.
Install
```
/skills add googleworkspace/cli --skill gws-workflow-email-to-task
```
Build HTML email as React components with react-email by resend. React Email, from Resend, including the visual editor integration. 19.7k★, MIT.
Install
```
/skills add resend/react-email --skill react-email
```
Stop your mail landing in spam with email-best-practices by resend. SPF, DKIM, DMARC, bounce handling, and the list hygiene part everyone skips. 140★.
Install
```
/skills add resend/email-best-practices --skill email-best-practices
```

Subagents and agent teams

Hand independent tasks to child sessions with subagent-driven-development by obra. Executes a plan by delegating the parts that do not share state. 279.9k★, MIT.
Install
```
/skills add obra/superpowers --skill subagent-driven-development
```
Fan work out when two or more tasks are independent with dispatching-parallel-agents by obra. The decision rule for when parallelism helps, and when it just costs tokens. 279.9k★, MIT.
Install
```
/skills add obra/superpowers --skill dispatching-parallel-agents
```
Keep the main context clean while investigating with research by warpdotdev. Pushes noisy searching into subagents and works from the distilled answer. 482★, MIT.
Install
```
/skills add warpdotdev/common-skills --skill research
```
Decide how many agents a job needs with team-composition-patterns by wshobson. Sizing heuristics and preset team shapes instead of spawning ten and hoping. 39.3k★, MIT.
Install
```
/skills add wshobson/agents --skill team-composition-patterns
```
Give a team of agents a message discipline with team-communication-protocols by wshobson. Message types, plan approval, shutdown, and the anti-patterns to avoid. 39.3k★, MIT.
Install
```
/skills add wshobson/agents --skill team-communication-protocols
```
Design durable multi-step workflows with workflow-orchestration-patterns by wshobson. Temporal-style separation of workflow and activity, sagas, and deterministic state. 39.3k★, MIT.
Install
```
/skills add wshobson/agents --skill workflow-orchestration-patterns
```

Build agents, skills, and MCP servers

Build an MCP server that models can actually use with mcp-builder by anthropics. Tool design, schemas, and error surfaces, from Anthropic. 172.7k★.
Install
```
/skills add anthropics/skills --skill mcp-builder
```
Build and debug TypeScript MCP servers with mcp-apps-builder by mcp-use. Tools, resources, prompts, middleware, and auth with mcp-use. 10.5k★, MIT.
Install
```
/skills add mcp-use/mcp-use --skill mcp-apps-builder
```
Scaffold a TypeScript MCP server with typescript-mcp-server-generator by github. A complete project on the official SDK, not a snippet. 38.5k★, MIT.
Install
```
/skills add github/awesome-copilot --skill typescript-mcp-server-generator
```
Scaffold a Python MCP server with python-mcp-server-generator by github. Same shape, Python SDK. 38.5k★, MIT.
Install
```
/skills add github/awesome-copilot --skill python-mcp-server-generator
```
Scaffold a Rust MCP server with rust-mcp-server-generator by github. Tools, prompts, resources, and tests on the official rmcp SDK. 38.5k★, MIT.
Install
```
/skills add github/awesome-copilot --skill rust-mcp-server-generator
```
Start an agent project with Google's ADK with google-agents-cli-scaffold by google. Project scaffold, with sibling skills for eval, deploy, and observability. 5.8k★, Apache-2.0.
Install
```
/skills add google/agents-cli --skill google-agents-cli-scaffold
```
Give the agent memory between sessions with agentmemory-mcp-tools by rohitg00. A map of every agentmemory tool and when to call which one. 27.8k★, Apache-2.0.
Install
```
/skills add rohitg00/agentmemory --skill agentmemory-mcp-tools
```

fx MCP servers

fx ships a full MCP client: Streamable HTTP, SSE, and stdio, with OAuth 2.1 including PKCE, dynamic client registration, and refresh. Remote servers are hand-added to ~/.fx/mcp.json, then /mcp reload. A literal Authorization header is refused by design, so a static key always goes through an environment variable.

The two config shapes, once

Delegated sign-in, no key to handle. Add the block, /mcp reload, then /mcp auth <name> --open:

{
  "mcp": {
    "linear": {
      "type": "http",
      "url": "https://mcp.linear.app/mcp"
    }
  }
}

Static key. Export it in your shell before starting fx; it never touches the file:

{
  "mcp": {
    "example": {
      "type": "http",
      "url": "https://mcp.example.com/mcp",
      "bearer_token_env": "EXAMPLE_API_KEY"
    }
  }
}

A local stdio server uses "type": "local" and an array command:

{
  "mcp": {
    "playwright": {
      "type": "local",
      "command": ["npx", "-y", "@playwright/mcp@latest"]
    }
  }
}

Every server here answered an MCP initialize with a proper RFC 9728 challenge, which is exactly the chain fx walks. Add the URL, /mcp reload, /mcp auth <name> --open.

GitHub. Repos, issues, pull requests, code search, and Actions. https://api.githubcopilot.com/mcp/
Linear. Issues, projects, cycles, and comments. https://mcp.linear.app/mcp
Notion. Pages, databases, and search across a workspace. https://mcp.notion.com/mcp
Sentry. Issues, events, and release health, plus Seer for root-cause analysis. https://mcp.sentry.dev/mcp
Supabase. Projects, tables, migrations, logs, and Edge Functions. https://mcp.supabase.com/mcp
Neon. Postgres projects, branches, and SQL against a branch. https://mcp.neon.tech/mcp
Vercel. Projects, deployments, logs, and environment variables. https://mcp.vercel.com
Cloudflare Workers Bindings. Create and inspect KV, R2, D1, and Workers from the session. https://bindings.mcp.cloudflare.com/mcp
Stripe. Customers, products, prices, invoices, and payment links. https://mcp.stripe.com
Atlassian. Jira issues and Confluence pages. https://mcp.atlassian.com/v1/sse
Asana. Tasks, projects, and portfolios. https://mcp.asana.com/sse
Canva. Designs, brand assets, and exports. https://mcp.canva.com/mcp
Intercom. Conversations, contacts, and help-centre articles. https://mcp.intercom.com/mcp
PayPal. Invoices, orders, subscriptions, and disputes. https://mcp.paypal.com/mcp
Square. Catalog, orders, payments, and customers. https://mcp.squareup.com/sse
Webflow. Sites, collections, and CMS items. https://mcp.webflow.com/sse
Wix. Site content, stores, and bookings. https://mcp.wix.com/sse
Apify. Five thousand scrapers as tools, and it takes a static token too. https://mcp.apify.com
Globalping. Ping, traceroute, DNS, and HTTP checks from a global probe network. https://mcp.globalping.dev/sse

No account needed

These answer an unauthenticated initialize. Add the URL and /mcp reload, nothing else.

Context7. Version-correct documentation and code examples for thousands of libraries, fetched at the moment the model needs them. https://mcp.context7.com/mcp
DeepWiki. Ask questions about any public GitHub repository and get answers grounded in its actual source. https://mcp.deepwiki.com/mcp
grep.app. Regex search across millions of public repositories, from Vercel. https://mcp.grep.app
Cloudflare Docs. Search Cloudflare's documentation without leaving the terminal. https://docs.mcp.cloudflare.com/mcp
Hugging Face. Search models, datasets, spaces, and papers. Anonymous works; a token raises the limits. https://huggingface.co/mcp
Exa. Neural web search built for agents. https://mcp.exa.ai/mcp

Run locally over stdio

{
  "mcp": {
    "chrome-devtools": {
      "type": "local",
      "command": ["npx", "-y", "chrome-devtools-mcp@latest"]
    }
  }
}

Chrome DevTools. Drive a real Chrome, read the DOM and console, capture performance traces. npx -y chrome-devtools-mcp@latest
Playwright. Browser automation over the accessibility tree rather than screenshots. npx -y @playwright/mcp@latest
Filesystem. Scoped read and write outside the workspace fx already sees. npx -y @modelcontextprotocol/server-filesystem /path
Memory. A knowledge graph that persists across sessions. npx -y @modelcontextprotocol/server-memory
Sequential Thinking. A structured scratchpad for problems that need steps. npx -y @modelcontextprotocol/server-sequential-thinking
Git. Read, search, and manipulate a repository as tools rather than shell calls. uvx mcp-server-git --repository /path

Live data APIs

Search YouTube and pull transcripts with transcriptapi by ZeroPointRepo. Transcripts, search, channels, and playlists. Free tier, no Google API key.
Config
```
{
  "mcp": {
    "transcriptapi": {
      "type": "http",
      "url": "https://transcriptapi.com/mcp",
      "bearer_token_env": "TRANSCRIPTAPI_KEY"
    }
  }
}
```
export TRANSCRIPTAPI_KEY=... before starting fx, then /mcp reload. Get a key at transcriptapi.com. It also answers an OAuth challenge, so /mcp auth transcriptapi --open works instead if you would rather not hold a key.
Look up US property data with zillapi by ZeroPointRepo. Listings, valuations, and property search. Free tier available.
Config
```
{
  "mcp": {
    "zillapi": {
      "type": "http",
      "url": "https://api.zillapi.com/mcp",
      "bearer_token_env": "ZILLAPI_API_KEY"
    }
  }
}
```
export ZILLAPI_API_KEY=... before starting fx, then /mcp reload. Get a key at zillapi.com.

Gateways and bridges

fx talks to Vercel AI Gateway, an OpenAI or ChatGPT subscription, or an xAI Grok subscription. Pointing it at anything else (llama.cpp, Ollama, vLLM, LM Studio, OpenRouter, LiteLLM) is the single most-solved problem in this ecosystem, by five separate people in under two weeks. Here is all of it in one place.

fx-any by redstar-systems. A ~230-line shim in front of the stock binary, using fx's own loopback gateway override. Survives fx upgrade because nothing is patched. Proven end to end against a local llama.cpp model, tool calls included. Apache-2.0.
fx-bridge by elijah7x. Python standard library only, no dependencies, loopback-only listener, HTTPS-only upstreams, explicit model mapping, and fail-closed routing. Ships a smoke test to run after every fx upgrade. Apache-2.0.
fx-lan by agent2x0r. One compiled Go binary that starts the local gateway and launches the real fx CLI for you. No runtime libraries. MIT.
fx-openai by iamdemetris. The fork rather than the shim: OPENAI_API_KEY and FX_OPENAI_BASE_URL are read natively, so there is no proxy in the path. You give up tracking upstream. Apache-2.0.
fx-gateway-proxy by Xeron2000. Comes at it from the other end: an OpenAI-compatible reverse proxy over Vercel AI Gateway's free GLM pool, with multi-key routing, learned capacity ceilings, and cooldown rotation. MIT. Archived by its author on 2026-08-29 — it still runs, but expect no fixes, and prefer one of the four above if you are starting fresh.

Ports and packaging

fx-windows by eng2007. Native x86_64-windows build through a compatibility layer, for the platform the official binary does not cover. Apache-2.0.
fxrs by Leeeon233. A compact Rust rewrite with a full-screen TUI and an ACP stdio server. The TUI is itself an ACP client, so editor integration and interactive use share one code path. Apache-2.0.
xfx by 2lab-ai. A Rust port of the load-bearing agent loop, with a parity ledger and CI that fails if the binary advertises anything the ledger does not record as implemented. Different binary name and profile home, so it cannot shadow an upstream install. Apache-2.0.
fx-ruby by jagenaujagenau. Ruby standard library only, zero gems, carrying over the system prompt, tool contracts, permission model, and session format. Apache-2.0.
homebrew-fx by Temikus. brew install temikus/fx/fx. Note the name clash: homebrew-core's fx is the JSON viewer, and only one can own the binary name.
fx-sandbox by da-beda. fxs runs the unmodified upstream binary inside a read-only container with all capabilities dropped, no Docker socket, and per-project state. It exits rather than silently falling back to native fx. MIT.
fx-sbx-kit by shelajev. A Docker Sandboxes kit, one sbx run --kit away, with host-managed Vercel OAuth or a gateway key.
faku by keejkrej. A native Zig desktop window for coding agents, fx-first, Waku-protocol compatible.
vercel-fx-cn by bbylw. The documentation in Chinese.

Embedding fx

WebAssembly SDK. Official. createFxAgent for headless ACP and createFxTerminal for the real TUI, both from sdk/ in the fx repo. Needs JSPI, so Chrome or Edge 137+.
ACP server. Official. fx acp speaks the Agent Client Protocol over stdio, which is how an editor drives fx as its backend.
chrome-fx by Lakphy. The official WASM terminal injected as a Chrome overlay, with browser and js commands so the agent drives the page you are looking at.
fx-on-workers by codingstark-dev. fx inside a Cloudflare Worker, one Durable Object per session holding a live wasm instance, TUI streamed over a WebSocket into xterm.js.
fx-in-a-tab by YuktiKholiwal. fx compiled to WebAssembly driving just-bash, with a real directory mounted read-only from your disk.
fx-review-action by TommyBez. AI review on every pull request. Validates each finding against the diff before posting, so hallucinated line numbers never reach the PR, and never blocks a merge. Use uses: TommyBez/fx-review-action@v0. The README still names an older account and a tag that does not exist.
fx-inside-function by NicolasMontone. fx running inside a serverless function, the smallest demonstration of the embedding story.

Good to know

How every entry is checked

Every skill on this page is re-checked weekly against the rule fx itself applies when it installs one, not against a link status.

flowchart LR
    A["Entry on this page\n/skills add owner/repo --skill name"] --> B["Resolve owner/repo\nthrough the GitHub API\ncatches renames and archives\na 200 status hides"]
    B --> C["Walk the default branch\nfor every SKILL.md\nthe same walk fx does"]
    C --> D{"A SKILL.md whose\nfolder name OR frontmatter\nname matches --skill?"}
    D -->|yes| E["Listed"]
    D -->|no| F["Flagged, never silently kept"]
    E --> G["Weekly re-check\nActions workflow, not memory"]
    G --> B

That is a real workflow in this repo (verify-skills.yml). It runs every Monday, and the two badges at the top of this page are written by that run. Neither is a hand-set claim: the workflow overwrites badges/verified.json every time, and the badge displays that file.

The match rule is taken from fx's own installer, which filters a cloned repository by folder name or by the name: in the SKILL.md frontmatter. A checker that only looked for the folder would report working entries as broken.

🛡️ Security notice

This is a curated list, not a security audit. A skill is a folder of instructions your agent will follow, and an MCP server is a process or endpoint acting with whatever authority you hand it. Both can read your files, spend your credits, and reach the network. A skill's presence here means it installs and the project is real and active as of its last check, not that its contents have been reviewed.

Read a SKILL.md before you install it, the same as you would a browser extension. Keep ~/.fx/mcp.json private, and use narrowly scoped tokens.

Found something malicious rather than merely broken? Open an issue and say so plainly, or use GitHub's private vulnerability reporting on the project's own repo.

🤝 Contributing

PRs are very welcome. CONTRIBUTING.md has the format and the acceptance rules.

Related lists

awesome-hermes-skills. Skills, plugins, memory providers, and surfaces for Nous Research's Hermes Agent.
awesome-dsh-plugins. Verified DeepSeek Harness plugins, organised by the job they do.
awesome-agent-plugins. The vendor-neutral Agent Plugins standard, and who actually complies with it.
awesome-cursor-plugins. Cursor plugins from the official marketplace, each with the other agents it also runs in and whether it needs a sign-in.
awesome-grok-bot. Skills, plugins, and MCP servers for Grok Bot.

Maintained by ZeroPointRepo · list content licensed CC BY 4.0 · Built with crhq.ai
_{Unofficial, community-maintained. Not affiliated with or endorsed by Vercel or Vercel Labs.}