✨ Free · takes 30 seconds

Claude Skill Generator

Turn any workflow into a Claude skill

Describe the task, pick your role, list the tools you use. Get a complete skill package — SKILL.md with trigger conditions, step-by-step process, and output format, plus reference files your agent consults mid-task — ready to install in Claude Code.

Be specific — mention tools, outputs, and the role this is for. The more detail, the better the skill.

Try an example:

What gets generated

📋

SKILL.md

Trigger conditions, required inputs, step-by-step process, and output format — specific to your exact workflow.

Reference files

Checklists, templates, lookup tables, and sequence blueprints tailored to the tools and domain.

📦

Ready to install

Download as a ZIP, drop it in ~/.claude/skills/, and it's active in your next session.

Already have skills you love? Browse 405+ verified skills or read the install guide.

What is a Claude skill?

A Claude skill is a plain-text instruction file — typically named SKILL.md — that you place in ~/.claude/skills/. Claude Code reads it at session start and follows the workflow inside whenever the matching task comes up. Skills are not plugins or extensions — they are just structured prompts that turn a repeatable job into a reliable, agent-ready process.

You can install pre-built skills from the directory, or use this generator to create a custom skill for a workflow that doesn't exist yet.

How to create a Claude skill

A production-ready skill has four sections: a name and activation line, trigger conditions (when to use it and when not to), required inputs the agent should confirm before starting, and a step-by-step process with an exact output format. Most skills also include reference files — checklists, templates, or lookup tables — that the agent consults mid-task.

The generator above handles all of this. Describe your workflow in plain English, select your role, list the tools you use (Apify, Clay, HubSpot, Notion, etc.), and the generator produces a complete package following the same structure used by the best skills in the directory.

For a deeper walkthrough, see the complete guide to building skills for Claude.

Claude skills examples

Skills work for almost any role and workflow. Some examples from the directory:

Browse all 405+ skills in the directory →

Claude skills best practices

Name real things

If the skill uses Apify, reference a specific actor (apify/linkedin-profile-scraper). If it uses Clay, name the exact enrichment columns (Find Work Email, Job Title). Generic instructions produce generic results.

List every required input

What would a consultant ask in the first five minutes? Every piece of context the agent needs — target audience, deal size, company stage, access to data sources — should be listed as a required input with "ask before proceeding" if missing.

Write process steps that cover conditionals

Vague steps ("research the prospect") fail in edge cases. Specific steps with branches ("if the LinkedIn URL is missing, fall back to company domain + hunter.io") make skills reliable.

Include reference files for lookup-heavy work

Skills for legal work, finance, and outreach almost always benefit from companion files: a clause library, a metrics definitions table, a sequence playbook. The generator creates these alongside SKILL.md automatically.

Keep the activation line narrow

A skill activated by "anything involving data" is almost never triggered correctly. "Activate when the user asks to build a weekly performance report from a CSV export" fires precisely when you want it to.

How to use Claude skills

Once installed, Claude skills are passive — you don't invoke them by name. Claude Code reads every SKILL.md in ~/.claude/skills/ at session start and applies whichever skill matches the task you describe. The trigger conditions in SKILL.md tell it when to activate.

Skills work with Claude Code, Claude.ai (via Projects), Codex CLI, Gemini CLI, and OpenClaw. For step-by-step setup instructions, see the install guide.

Frequently asked questions

What is a Claude skill?

A Claude skill is a plain-text instruction file (SKILL.md) placed in ~/.claude/skills/. Claude Code reads it at session start and follows the workflow inside whenever the matching task comes up. Skills define trigger conditions, required inputs, a step-by-step process, and an exact output format — turning any repeatable job into a reliable, agent-ready process.

How do I create a Claude skill?

A production-ready skill needs four sections: a name and activation line, trigger conditions, required inputs, and a step-by-step process with output format. Most skills also include reference files — checklists, templates, or lookup tables. Use the generator above: describe your workflow, select your role, list the tools you use, and download the ZIP.

How do I use Claude skills after installing them?

Skills are passive once installed — you don't invoke them by name. Claude Code reads every SKILL.md at session start and applies whichever skill matches the task you describe. The trigger conditions in the skill tell it when to activate. Skills work with Claude Code, Claude.ai (via Projects), Codex CLI, Gemini CLI, and OpenClaw.

What is the difference between a Claude skill and a custom instruction?

Custom instructions apply globally to every conversation and shape tone or general behavior. Claude skills are task-specific — each one activates only when a matching workflow is detected, and carries a full process definition, required inputs, and output format. Skills are more precise, more maintainable, and can include companion reference files the agent consults mid-task.

What does the Claude Skill Generator produce?

A complete skill package: a SKILL.md file with trigger conditions, required inputs, step-by-step process, and output format — plus 1–3 reference files (checklists, templates, lookup tables) tailored to your tools and domain. Everything is packaged as a downloadable ZIP, ready to drop into ~/.claude/skills/.