Skip to content

First Run

1. Install skillx

If you have not installed skillx yet, use the recommended install command:

Terminal window
curl -fsSL https://skillx.run/install.sh | sh

2. Run a Real Agent Skill

Use the same public Agent Skill featured on the landing page:

Terminal window
skillx run https://github.com/anthropics/skills/tree/main/skills/frontend-design "Redesign the hero section."

3. Expect This Kind of Output

You should see a flow like this:

● Resolving source...
✓ Resolved: frontend-design
● Scanning for security issues...
✓ PASS - no findings
● Injecting skill...
✓ Injected 1 files
● Launching agent...

4. What skillx Is Doing

skillx fetches the skill from GitHub, scans it before any injection happens, copies the skill into the right agent-specific location, launches your agent with the skill in context, and then cleans up the temporary files when the run finishes.

This is a stronger first run than a toy example: you get a visible result from a real upstream skill without committing to a permanent install.

That sequence is the point of the tool: one command to run a skill safely without leaving a permanent install behind.

Next Steps

  • Famous Skills - try the most useful skills next
  • Official Examples - browse the full set of runnable examples
  • Run Skills - learn the command in detail
  • FAQ - understand when to stay in one-off mode and when to move deeper
  • Troubleshooting - debug source resolution, agent detection, and scan gate failures