Merging Slash Commands into Skills in Claude Code
Claude Code has merged Slash Commands into Skills. No migration is needed - existing slash commands in ~/.claude/commands continue to work as normal.
Why Combine Slash Commands and Skills?
Slash Commands were one of the first abstractions for managing context, serving as a form of progressive-disclosure. They ensured the model only loaded context when needed.
As model capabilities advanced, Skills proved to be the more powerful approach:
- Dynamic context loading - Skills allow the model to load context dynamically by reading relevant files
- Multi-level context - You can reference other files inside your SKILL.MD for multiple levels of dynamic context
- Simplified mental model - One abstraction instead of two (no separate SlashCommand Tool and Skill Tool)
Invocation Control
For each skill, you can choose whether it should be:
- User-invocable - Can be invoked with slash command syntax (
/skillname) - Model-invocable - Can be called automatically by Claude
- Both (the default)
# Prevent user invocation
user-invocable: false
# Prevent model invocation
disable-model-invocation: true
Using Subagents with Skills
Skills naturally pair with subagents. Subagents allow you to execute skills while protecting your context window.
Search Skills with the Explore Agent
Setting agent: <agent-name> spawns a subagent that loads the skill into its context.
Example: A Research skill that summarizes files using the Explore agent and returns the result.
Memory Skills with Forked Context
Setting context: fork spins off a subagent with all your current context. Great for parallel operations.
Example: A memory skill to summarize your conversation and save it to a file - these tool results stay out of your main context since they're unrelated to the current work.