Skills¶
Skills are reusable capabilities for AI agents. Install them with a single command to enhance your agents with access to procedural knowledge.
Install Skills¶
Example how to setup skills with subagents and scripts, etc.
| Environment | Path | Writable? | What's There |
|---|---|---|---|
| Cowork skills | .skills/skills/ |
NO (read-only during sessions) | Installed Cowork skills |
| Cowork plugins | .local-plugins/marketplaces/local-desktop-app-uploads/ |
NO (read-only) | Multi-skill plugins |
| Remote plugins | .remote-plugins/ |
NO (read-only) | Marketplace plugins |
| Claude Code | .claude/skills/ |
YES | All skills as folders |
| Cursor | .cursor/skills/ |
YES | |
| Source |
Just Some Skills¶
Socrates Skill¶
Works well with code or documents.
Never answers. Always asks. A Socratic method teaching skill that guides you to discover answers through questioning — for any knowledge asset.
https://github.com/bevibing/socrates-skill
Socratic Method Skill for Claude¶
A more complex and maybe deeper approach to the Socratic method.
This skill implements the classical Socratic method (maieutics) for pedagogical interactions, cognitive restructuring, strategic coaching, and analytical thinking. Instead of delivering direct answers, Claude uses disciplined questioning to help users develop deep understanding through self-discovery.
https://github.com/malkreide/socratic-method-skill
Invoking Skills¶
Claude Code doesn't have a built-in disambiguation mechanism for this — it relies on how skills are referenced and described.
Here's how it actually resolves which one to use:
By exact name match in the slash command
If you invoke /socrates, Claude Code looks for a skill whose folder name matches exactly. So:
/socrates→.claude/skills/socrates/SKILL.md/socrates-method→.claude/skills/socrates-method/SKILL.md
The folder name is the command name.
By description matching (when invoked ambiguously)
If you trigger a skill via natural language rather than a slash command, Claude reads the description field in each SKILL.md and picks the one whose description best fits your request. This is where conflicts can arise — if both descriptions are similar, it may pick the wrong one or ask you to clarify.
Practical recommendations
- Keep folder names distinct and descriptive (
socratesvssocrates-methodis already fine) - Make sure the
descriptionfields in the two SKILL.md files are clearly differentiated — they should describe different use cases or trigger conditions so there's no overlap - If you want one to be the "default" Socratic skill and the other more specialized, make that explicit in the descriptions
So in your case: no conflict for slash command invocation since the names differ. The only risk is if both descriptions are so similar that natural-language triggering becomes ambiguous.