Getting Started

Go from zero to a connected IDE in under five minutes. This guide walks through account creation, access key generation, and your first skill discovery call.

On this page

What is Skills Hub?

Skills Hub is a registry for AI agent skills built on the open agent-skills standard. It provides a single source of truth where teams can publish, version, and manage the skills that power their AI coding assistants.

Your IDE connects directly to the registry API so agents always receive the latest version of every skill -- no manual syncing, no stale files on disk.

Key concepts

Skill: A self-contained unit of capability defined by a SKILL.md file. Each skill has YAML frontmatter (metadata) and a markdown body (instructions).
Owner / Namespace: Every skill is scoped to an owner: anthropic/code-review, acme/api-docs. This prevents naming conflicts.
Progressive Disclosure: Discovery is lightweight (~100 tokens). Full SKILL.md loads only on activation (<5,000 tokens). Supporting files stream on-demand during execution.

Creating an Account

Head to skillshub.dev/signup and create an account with your email or sign in with GitHub. The free tier includes 5 private skills and 1,000 MCP calls per day -- plenty for getting started.

Plan	Private Skills	MCP Calls / Day
Free	5	1,000
Business	50	50,000
Enterprise	Unlimited	Unlimited

Generating an Access Key

After signing in, navigate to Settings → Access Keys in the dashboard. Click Create Key, give it a descriptive name (e.g. “cursor-work-laptop”), and copy the generated key.

Keep your key secret

Access keys start with sk_live_ and grant full read/write access to your account. Never commit keys to source control. Regenerate immediately if compromised.

You can have multiple keys active at the same time, one per machine or environment, and revoke individual keys without affecting others.

Connecting Your IDE

Skills Hub exposes an MCP (Model Context Protocol) server that any compatible IDE can connect to over HTTP. No local process or CLI needed — just a URL and your access key.

Claude Code

Run this command in your terminal:

claude mcp add --transport http skillshub \
  https://skillshub.dev/api/mcp \
  --header "Authorization: Bearer sk_live_your_access_key_here"

Then restart Claude Code and verify with /mcp. Skills are automatically available as tools.

Cursor

Add to your project's .cursor/mcp.json:

{
  "mcpServers": {
    "skillshub": {
      "url": "https://skillshub.dev/api/mcp",
      "headers": {
        "Authorization": "Bearer sk_live_your_access_key_here"
      }
    }
  }
}

Windsurf

Add to ~/.codeium/windsurf/mcp_config.json:

{
  "mcpServers": {
    "skillshub": {
      "serverUrl": "https://skillshub.dev/api/mcp",
      "headers": {
        "Authorization": "Bearer sk_live_your_access_key_here"
      }
    }
  }
}

Verify the connection

After connecting, your IDE should be able to call Skills Hub MCP tools. In Claude Code, try running /mcp to see the available tools: discover_skills, search_skills, get_skill, get_skill_content, get_skill_file, and publish_skill.

Browsing and Discovering Skills

There are two ways to find skills:

1. The Explore page

Visit skillshub.dev/explore to browse all public skills. Filter by category, sort by popularity, and view skill details before using them in your agent.

2. MCP Discovery

Your IDE automatically calls the discover_skills MCP tool at startup to receive a lightweight list of available skills. Each entry includes only the fields an agent needs to decide relevance — roughly 100 tokens per skill.

For targeted lookups, your agent uses the search_skills tool with a keyword query:

// Your agent calls these MCP tools automatically:
discover_skills()           // List all available skills
search_skills("code review") // Search by keyword

See the MCP Reference for full tool documentation.

Understanding Skill Structure

Every skill is defined by a SKILL.md file that combines YAML frontmatter with markdown instructions. Here is a minimal example:

---
name: code-review
description: >
  Review code for bugs, security issues,
  and best practices.
version: 1.0.0
license: MIT
compatibility:
  - cursor
  - claude-code
  - jetbrains-ai
metadata:
  category: development
  tags:
    - code-quality
    - review
allowed-tools:
  - Read
  - Grep
  - Glob
---

# Code Review

You are an expert code reviewer. When activated,
review the provided code for:

1. **Bugs** - logic errors, off-by-one, null refs
2. **Security** - injection, auth issues, secrets
3. **Best practices** - naming, structure, DRY

Provide clear, actionable feedback with code
examples for each finding.

The frontmatter (between the --- markers) contains structured metadata the registry uses for indexing, discovery, and compatibility filtering.

The markdown body below the frontmatter is the actual instruction content that gets injected into the agent's context when the skill is activated.

Skills can also include supporting files in scripts/, references/, and assets/ directories. These stream on-demand during execution to keep initial context small.

Next steps

Now that your IDE is connected, learn how to create and publish your own skills.

Publishing Guide MCP Reference