
vercel-labs/skills
find-skills
Lets an agent search the open skills ecosystem when a task calls for specialized knowledge.
Cody skill library
A researched collection of useful agent skills for frontend work, marketing, workflows, MCP, RAG, databases, documents, and legal writing — with source links, audit snapshots, and honest caveats.
27
hand-picked skills
18
first-party picks
3.2M
top snapshot installs
Install counts and audits are point-in-time signals, not a guarantee. Read every SKILL.md, inspect scripts and network behavior, and start with the smallest set that fits your work. Snapshot checked September 1, 2026.
27 skills

vercel-labs/skills
Lets an agent search the open skills ecosystem when a task calls for specialized knowledge.

anthropics/skills
Gives an agent a stronger visual point of view before it starts writing interface code.

vercel-labs/agent-browser
Adds persistent, command-driven browser automation for navigation, extraction, form work, and web QA.

vercel-labs/agent-skills
A prioritized React and Next.js performance checklist maintained by Vercel Engineering.

microsoft/azure-skills
First-party operational guidance for building, evaluating, deploying, and managing Microsoft Foundry agents and models.

supabase/agent-skills
A prioritized Postgres playbook for schema design, security, query performance, and operations.

obra/superpowers
Keeps an agent in a root-cause investigation loop before it starts patching symptoms.

coreyhaines31/marketingskills
Turns a broad SEO review into a prioritized diagnosis across technical, on-page, and content signals.

coreyhaines31/marketingskills
A conversion-copy workflow for pages, headlines, value propositions, and calls to action.

anthropics/skills
A design and implementation guide for MCP servers that expose external services to language models.

anthropics/skills
Structures substantial writing as context transfer, iterative drafting, and fresh-reader testing.

google/skills
Manages Google Agent Platform RAG corpora, files, retrieval, and grounded generation through the Gen AI SDK.

anthropics/claude-for-legal
Gives structural feedback on legal writing without silently replacing the author's analysis or voice.

mattpocock/skills
Guides vertical, behavior-focused red-green-refactor cycles instead of writing a large test suite up front.

vercel-labs/agent-skills
Reviews interface code against a maintained set of usability, interaction, and accessibility checks.

mattpocock/skills
A diagnosis loop for stubborn failures and performance regressions where the cause is not yet known.

remotion-dev/skills
Routes an agent through the right Remotion guidance for building video in React.

anthropics/skills
A full workflow for drafting, testing, measuring, and improving an agent skill.

vercel-labs/agent-skills
Helps agents design React component APIs that remain flexible as a product grows.

prisma/skills
First-party setup and operations guidance for Prisma Postgres across console, CLI, and management APIs.

anthropics/skills
Creates, reads, edits, and validates professional Word documents with formatting-aware workflows.

coreyhaines31/marketingskills
Applies behavioral science and decision models to messaging, pricing, and conversion choices.

coreyhaines31/marketingskills
Builds a content plan around audience demand, business goals, and a balanced portfolio of ideas.

coreyhaines31/marketingskills
Plans scalable search pages around repeatable query patterns, useful data, and sound indexation controls.

coreyhaines31/marketingskills
Evaluates whether content is easy for answer engines and AI assistants to extract, trust, and cite.

anthropics/skills
A current reference skill for Claude models, API patterns, tools, caching, streaming, MCP, and migration.

google/skills
Google's implementation guide for Gemini through Vertex AI and Agent Platform.
How the list was made
A deliberately small set of useful, inspectable agent skills selected from current skills.sh usage, first-party repositories, GitHub activity, independent roundups, and practitioner discussion.
Prefer focused skills that encode a repeatable workflow or a maintained domain playbook.
Give first-party repositories extra weight, then use installs, repository activity, and community discussion as supporting signals.
Show audit warnings and low-adoption exceptions instead of treating popularity as a security guarantee.
Keep the collection small enough that users can install only what matches their work.
Browse the rebuilt prompt library for detailed, model-aware templates designed for GPT-5.6 and Claude Fable 5.1.
Explore prompt templates