Claude Code

Explain does the most from an agent that can read your code. Connected to Scrimba, Claude Code doesn't ask someone else to explain something. It writes the explainer itself, with your actual repository in front of it. That means explainers of things no general model could know:

›explain what this branch changes and why›walk me through how authentication works in this codebase›make an explainer of this PR for the reviewer›explain the bug we just fixed and how

Connect with the magic link

Add Explain to ClaudeOpens Claude with the Explain connector prefilled, ready to confirm.→

Once you confirm, the connector is on your account and available in Claude Code and on claude.ai.

Or connect with a command

claude mcp add --transport http scrimba-explain https://scrimba.com/explain/mcp

Either way, from your next session Claude Code can create explainers whenever you ask.

How it works

When you ask for an explainer, Claude Code plans the lesson, then streams it to Scrimba slide by slide: narration script, code excerpts, diffs, diagrams. You get a link as soon as the stream starts.

The draft it creates is private: nobody can stumble onto it, and the link alone isn't enough to watch it. Opening the link signs you in to Scrimba and claims the explainer as yours. From that point it's in your account and you decide who else can see it. See Privacy, claiming and sharing.

Creating the draft is free: Claude Code authors the lesson, so no Scrimba AI generation is involved. Narration and images are generated when you claim it and watch.

Other agents

Codex and every other MCP-capable tool can connect to the same server. See The MCP server.