Loading lesson…
Skills let you package a prompt, tools, files, and configuration into a named capability Claude can invoke on demand.
Skills are Anthropic's 2025 answer to the question 'how do I teach Claude to do a thing really well, once, and have it work every time?' A Skill is a directory containing a SKILL.md definition file, optional helper scripts, and any reference files Claude needs.
--- name: weekly-report trigger: When the user asks for a weekly status report --- # Weekly Report Skill When invoked: 1. Read the last 7 days of entries from `journal/` using `tools/read-journal.sh` 2. Summarize into 5 bullets: wins, struggles, surprises, next-week, ask. 3. Write output to `out/report-YYYY-MM-DD.md` 4. Return the path and a 1-line summary.A minimal SKILL.md| Concept | What it is | Use when |
|---|---|---|
| Claude Project | A persistent chat context with shared files | Repeated conversations on the same topic |
| Skill | Packaged expertise — invoke on demand | Specialized tasks with well-defined I/O |
| MCP Server | A tool-provider Claude can connect to | Integrating external systems (DBs, APIs, file systems) |
Skills run inside the Claude environment. MCP servers run outside and expose tools. Projects are Claude's working memory. You often use all three.
8 questions · take it digitally for instant feedback at tendril.neural-forge.io/learn/quiz/end-latest-claude-skills-creators
What is the main idea of "Claude Skills — reusable specialized agents"?
Which concept is most central to "Claude Skills — reusable specialized agents"?
Which use of AI fits this topic best?
What should a careful learner remember about "Scope your agents tightly"?
You want to use AI after this lesson. What is the safest next step?
How should AI output about Claude Skills be treated?
Name one way to verify an AI answer about Claude Skills.
Which action would help you apply "Claude Skills — reusable specialized agents" responsibly?