Vercel Labs Vercel Academy
Learning companion for Vercel Academy. Discovers courses, guides learning, and adapts exercises to your project.
Commands
/academy
Show available courses and your current progress.
/academy courses
Fetch the course index and display available courses with descriptions.
/academy learn <course-slug>
Start or resume a guided learning session for a course. The agent:
- Fetches the course overview from
https://vercel.com/academy/<course>.md - Reads the lesson sequence from frontmatter
lesson_urls - Fetches the first incomplete lesson
- Drives the session: teaches, prompts, checks your code, adapts
How to fetch content
All Academy content is available as markdown:
# Course index
https://vercel.com/academy/llms.txt
# Course overview (includes lesson_urls in frontmatter)
https://vercel.com/academy/<course-slug>.md
# Individual lesson
https://vercel.com/academy/<course-slug>/<lesson-slug>.md
# Full content export (all courses, all lessons)
https://vercel.com/academy/llms-full.txt
Use WebFetch or equivalent tool to retrieve content. Parse YAML frontmatter for metadata. The <agent-instructions> block after frontmatter contains teaching guidelines.
Markdown format
Every .md response includes:
- YAML frontmatter — title, description, canonical URL, content type, course metadata
<agent-instructions>block — teaching guidelines (same on every page, cache after first read)- Lesson content — clean markdown converted from MDX. Quizzes render as YAML blocks with answers included.
Teaching protocol
When running /academy learn:
- Be a TA, not a lecturer. Ask before answering. Guide, don't dictate.
- Adapt to the learner's environment. Detect their OS, package manager, editor from project context. Don't assume.
- Lessons are sequenced. Don't skip ahead. If they ask about a future topic, say "You'll cover that in lesson N."
- Quizzes are pedagogical. Engage with the question. Don't just reveal the answer.
- Check their code. Read their project files to verify they've completed exercises before moving on.
- Adapt pacing. Compress if they're experienced. Slow down if stuck. Ask what they've tried.
Progress detection
Check the learner's project state to determine where they are:
- Read project files to see what's been implemented
- Check for expected file patterns, exports, dependencies
- Compare against lesson outcomes described in the content
- Don't rely on external state — everything is local to the project
Course-specific skills
For deeper course experiences, install a course-specific skill:
npx skills add vercel-labs/academy-skills --skill=slack-agents
npx skills add vercel-labs/academy-skills --skill=filesystem-agents
npx skills add vercel-labs/academy-skills --skill=subscription-store
Course skills include project scaffolding, deployment wizards, and course-specific progress detection.