VixlVixl
Docs
Changelog
GitHub
Docs
Changelog
GitHub
  • Get Started

    • Overview
    • Philosophy
    • Installation
    • Set up providers and models
    • Add a project
    • Create a project group
    • Your first chat
  • Concepts

    • How Vixl works
    • Chat modes
    • Permissions and approvals
    • Projects and home chats
    • Models and roles
    • Context
    • Code graphs
    • The .vixl directory
  • Using Vixl

    • Manage chats
    • Queue and stop messages
    • Review and restore changes
    • Compact and hand off long chats
    • Export a transcript
    • Work with plans
    • Orchestrate sub-agents
    • Best practices
    • Use the workbench
    • Shortcuts and the command palette
  • Customize

    • Providers
    • Models
    • MCP servers
    • Skills
    • Custom agents
    • Rules and AGENTS.md
    • Language servers
    • Appearance
    • Permission settings
  • Reference

    • settings.json
    • mcp.json
    • Custom agent frontmatter
    • SKILL.md format
    • Keyboard shortcuts
    • Chat statuses
    • .vixl layout
    • Managed components
  • Resources

    • Troubleshooting
    • FAQ
    • Privacy
    • Comparison
    • Roadmap
    • Brand and press
    • Changelog

Skills

A skill is a folder with a SKILL.md file: a short catalog line plus a procedure the agent loads when it needs it. Use one for a repeatable workflow (deploy, a review checklist, release notes) so the full text is not in every prompt.

The complete field list is on SKILL.md format.

Create a skill

The fastest path is the vendored /create-skill command. Type / in the chat input, pick create-skill, and send. The agent writes the file. /create-agent, /create-rule, and /create-plan work the same way for those other files.

You can also create one from Settings > Skills (personal) or the project Skills tab. The form asks for a name, a description, and a markdown body, then writes the file. Click a row to open it in the workbench editor.

ScopeWhere you create itFile
PersonalSettings > Skills~/.vixl/skills/{slug}/SKILL.md
ProjectProject Skills tab<repo>/.vixl/skills/{slug}/SKILL.md

On a home chat, /create-skill writes ~/.vixl/skills/{slug}/SKILL.md. That path is the personal tree, the same files Settings > Skills edits. Paths are on the .vixl directory.

Name and description are required. The create form writes them as JSON strings. The body is markdown. Loaders strip the frontmatter and inject the body. Bodies over 4000 characters are truncated when loaded.

---
name: "deploy"
description: "Ship this app when the user asks to deploy"
---

1. Run the test suite.
2. Build the production bundle.
3. Push the release tag the user named.

Every field is documented on SKILL.md format.

How it takes effect

/ lists skills and custom agents. Selecting a skill inserts a mention. On send, that becomes a Skill {name} line in the prompt. The agent then calls load_skill with that name. It can also call load_skill from the catalog without a / mention.

Reserved names ask, plan, agent, and orchestrator cannot run via /. Those are chat modes. Their built-in skills are inlined only in the matching mode, then omitted from the catalog.

/create-skill, /create-agent, /create-rule, and /create-plan are vendored command skills. They show up in / and in Available skills, including on home chats.

Discovery and precedence

Same name, case-insensitive: a project skill overlays a personal one. Home chats, and a workspace rooted at your user home directory, use personal skills only. load_skill name resolution, / merge order, and vendored-command protection are on SKILL.md format. How catalogs land in the prompt is on Context.

Edit this page
Last Updated: 10/6/26, 8:58 PM
Prev
MCP servers
Next
Custom agents