Skip to content
SYN4PSE home

Terminal

Build your harness.

Skills, MCP servers and plugins from the companies and individuals that make them, each with the one line that installs it in the tools you already use.

  • Claude Code
  • Codex CLI
  • Cursor
  • Antigravity
  • Copilot CLI
  • opencode

6635 entries · 3 types · 6 tools

6635 entries

brainstorming

Helps your agent ask the right questions and agree on an approach and design doc with you before it builds anything new.

Skill

305kGitHub stars

404kinstalls through skills.sh

Works with: Claude Code, Codex CLI, Cursor, Antigravity, Copilot CLI, opencode.

systematic-debugging

Makes your agent find the root cause of a bug, test failure or unexpected behavior before it proposes any fix.

Skill

305kGitHub stars

306kinstalls through skills.sh

Works with: Claude Code, Codex CLI, Cursor, Antigravity, Copilot CLI, opencode.

writing-plans

Turns a spec for a multi-step task into a plan of small tasks with exact files, interfaces and tests, before any code is touched.

Skill

305kGitHub stars

283kinstalls through skills.sh

Works with: Claude Code, Codex CLI, Cursor, Antigravity, Copilot CLI, opencode.

using-superpowers

Tells your agent to check for and invoke any relevant skill before responding, and how to pick between skills when several apply.

Skill

305kGitHub stars

271kinstalls through skills.sh

Works with: Claude Code, Codex CLI, Cursor, Antigravity, Copilot CLI, opencode.

requesting-code-review

Has your agent send finished work to a fresh reviewer subagent, with commit range and requirements, to catch issues before merging.

Skill

305kGitHub stars

262kinstalls through skills.sh

Works with: Claude Code, Codex CLI, Cursor, Antigravity, Copilot CLI, opencode.

test-driven-development

Has your agent write a failing test first, watch it fail, then write the minimal code to pass, for features and bug fixes.

Skill

305kGitHub stars

259kinstalls through skills.sh

Works with: Claude Code, Codex CLI, Cursor, Antigravity, Copilot CLI, opencode.

verification-before-completion

Stops your agent from claiming work is done or passing until it has run the verification commands and read their output.

Skill

305kGitHub stars

251kinstalls through skills.sh

Works with: Claude Code, Codex CLI, Cursor, Antigravity, Copilot CLI, opencode.

executing-plans

Has your agent carry out an implementation plan itself in one session, task by task, with tests for each step and one final review.

Skill

305kGitHub stars

243kinstalls through skills.sh

Works with: Claude Code, Codex CLI, Cursor, Antigravity, Copilot CLI, opencode.

subagent-driven-development

Runs an implementation plan by giving each task to a fresh subagent, reviewing every task, then doing a broad review of the whole branch.

Skill

305kGitHub stars

238kinstalls through skills.sh

Works with: Claude Code, Codex CLI, Cursor, Antigravity, Copilot CLI, opencode.

receiving-code-review

Makes your agent check review feedback against the codebase and push back when it is wrong, instead of agreeing and implementing blindly.

Skill

305kGitHub stars

222kinstalls through skills.sh

Works with: Claude Code, Codex CLI, Cursor, Antigravity, Copilot CLI, opencode.

dispatching-parallel-agents

Teaches your agent to split independent problems, such as unrelated test failures, across separate subagents that work at the same time.

Skill

305kGitHub stars

218kinstalls through skills.sh

Works with: Claude Code, Codex CLI, Cursor, Antigravity, Copilot CLI, opencode.

writing-skills

Guides you in creating and editing agent skills by testing them with pressure scenarios on subagents before you deploy them.

Skill

305kGitHub stars

218kinstalls through skills.sh

Works with: Claude Code, Codex CLI, Cursor, Antigravity, Copilot CLI, opencode.

using-git-worktrees

Sets up an isolated workspace for feature work, using your agent's native worktree tools or a git worktree when none exist.

Skill

305kGitHub stars

217kinstalls through skills.sh

Works with: Claude Code, Codex CLI, Cursor, Antigravity, Copilot CLI, opencode.

finishing-a-development-branch

Guides your agent through finishing a branch: run the full tests, offer ways to integrate the work, then clean up the workspace.

Skill

305kGitHub stars

215kinstalls through skills.sh

Works with: Claude Code, Codex CLI, Cursor, Antigravity, Copilot CLI, opencode.

diagnosing-superpowers

Reads session transcripts to show what went wrong in a Superpowers session, with cited evidence, and can prepare a bug report for its maintainers.

Skill

305kGitHub stars

20kinstalls through skills.sh

Works with: Claude Code, Codex CLI, Cursor, Antigravity, Copilot CLI, opencode.

superpowers

Core skills library for Claude Code: TDD, debugging, collaboration patterns, and proven techniques

Plugin

305kGitHub stars

Works with: Claude Code, Codex CLI, Cursor, Copilot CLI.

grill-me

A relentless interview to sharpen a plan or design.

Skill

302kGitHub stars

1.4minstalls through skills.sh

Works with: Claude Code, Codex CLI, Cursor, Antigravity, Copilot CLI, opencode.

grill-with-docs

Interviews you relentlessly to sharpen a plan or design, and writes architecture decision records and a glossary as the answers come in.

Skill

302kGitHub stars

1.2minstalls through skills.sh

Works with: Claude Code, Codex CLI, Cursor, Antigravity, Copilot CLI, opencode.

improve-codebase-architecture

Scans your codebase for ways to make modules deeper, shows them in a visual HTML report, then interviews you through the one you pick.

Skill

302kGitHub stars

1.2minstalls through skills.sh

Works with: Claude Code, Codex CLI, Cursor, Antigravity, Copilot CLI, opencode.

tdd

Guides your agent to build features or fix bugs test-first with red-green-refactor, one small slice at a time.

Skill

302kGitHub stars

1.1minstalls through skills.sh

Works with: Claude Code, Codex CLI, Cursor, Antigravity, Copilot CLI, opencode.

setup-matt-pocock-skills

Configures your repo for Matt Pocock's engineering skills by setting up the issue tracker, triage labels and where domain docs live.

Skill

302kGitHub stars

1.1minstalls through skills.sh

Works with: Claude Code, Codex CLI, Cursor, Antigravity, Copilot CLI, opencode.

handoff

Compact the current conversation into a handoff document for another agent to pick up.

Skill

302kGitHub stars

1minstalls through skills.sh

Works with: Claude Code, Codex CLI, Cursor, Antigravity, Copilot CLI, opencode.

prototype

Builds a throwaway prototype to answer one design question, such as whether a state model feels right or what a UI should look like.

Skill

302kGitHub stars

988kinstalls through skills.sh

Works with: Claude Code, Codex CLI, Cursor, Antigravity, Copilot CLI, opencode.

domain-modeling

Helps you sharpen a project's domain language by challenging fuzzy terms, keeping a glossary file, and recording hard decisions as architecture decision records.

Skill

302kGitHub stars

878kinstalls through skills.sh

Works with: Claude Code, Codex CLI, Cursor, Antigravity, Copilot CLI, opencode.

teach

Teach the user a new skill or concept, within this workspace.

Skill

302kGitHub stars

862kinstalls through skills.sh

Works with: Claude Code, Codex CLI, Cursor, Antigravity, Copilot CLI, opencode.

What these are

Three things you can add to an agent.

Every entry is one of these. Pick one to see the shape it takes.

A folder with a SKILL.md. Your agent reads its name and description, and opens the full steps when a task matches.

5735 skills in the index

How you install

One line in your terminal.

Every entry opens on this panel. Pick where it goes and copy the command. Or copy the prompt, and your agent runs it for you.

  1. 01

    Where it goes

    Project puts it in the repository, so you commit it with your code. Global puts it in your home folder, so every project gets it.

  2. 02

    Command or prompt

    The command runs in your terminal. The prompt holds the same command for your agent to run.

  3. 03

    Run it

    The installer finds the tools you have and writes the files each one reads.

The entry shownnext-dev-loop

Install

Command or prompt
Where it installs
npx skills add vercel/next.js --skill next-dev-loop
A recorded run of the skills installer.

Tags

Read the tags before you run it.

Code sets each tag from the source. They say where an entry runs and what it can reach.

  • Runs on your computer

    It runs as a program on your machine, with your access.

    5846 entries

  • Runs in the cloud

    It runs on the publisher's servers. What you send it leaves your machine.

    433 entries

  • Needs a key or account

    It asks for an API key or a sign in before it works.

    417 entries

  • Runs shell commands

    Its steps run shell commands in your terminal.

    543 entries

  • Security audits

    Skills also show the scans of skills.sh's security partners. Each partner scans on its own schedule.

    • PassNothing flagged.
    • WarnSomething to read first.
    • FailA problem flagged.

Questions

Questions and answers.

Something else?

What is the difference between a skill, an MCP server and a plugin?

A skill is a set of instructions: a folder with a SKILL.md that tells your agent how to do one job. An MCP server is a connection: a program that gives your agent tools to read from or act in another app. A plugin is a bundle: skills, agents, commands, hooks and servers that install together.

Should I pick Project or Global?

Project writes the files into the repository you are in, so the people you work with get them when they pull. Global writes them into your home folder, so every project on your machine uses them.

Where do the files go?

A skill goes into .agents/skills/<name>, with a link in .claude/skills for Claude Code. An MCP server becomes one entry in each tool's config file, such as .mcp.json or .cursor/mcp.json. A plugin goes into each tool's plugin store in your home folder.

Which tools does it work with?

Claude Code, Codex CLI, Cursor, Antigravity, Copilot CLI and opencode. Each entry page greys the tools its installer does not reach in the scope you pick, and a * marks a tool with a note.

Who publishes these entries?

The companies and individuals that make them. An entry is listed only when its source sits in its publisher's own GitHub account or on its own website. An individual is listed once one of their repositories passes 25,000 GitHub stars.

Is it safe to install?

An entry runs with the access you give your agent. Read its tags and its source before you run it. Skills also show the security audits from skills.sh's partners.

How do I remove something I installed?

Run npx skills remove <name> for a skill, or npx add-mcp remove <name> for an MCP server. Remove a plugin with your tool's own plugin manager, such as /plugin in Claude Code.

Does it cost anything?

No. The index and every install line are free to use. Some entries need an account with their publisher.

How do I suggest an entry?

a link to its repository or docs. We list it when it comes from the company or individual that makes it.

Suggest

Missing something you use?

Send a link to its repository or docs. We list it when it comes from the company or individual that makes it.

Type
Optional
0 / 500
Optional