Claude Code Skill Map
Interactive game that tests how well you know Claude Code
Start new thread
Krishna Goyal

Claude Code Skill Map - Interactive game that tests how well you know Claude Code

by
What YC Is Really Betting On?
•
Claude Code just crossed 29M daily installs but most developers are using maybe 20% of what it can do. I built an interactive game covering: hooks, subagents, MCP servers, CLAUDE.md, slash commands and more. 6 rounds. 15 challenges. 5 different formats including a timed speed round. You get a spider-web radar chart across 7 skill areas and an archetype title based on your score. Some surprises: features most daily users had no idea existed. Free. No login.

Add a comment

Replies

Best
Krishna Goyal
Maker
📌
Hey PH 👋 I kept using Claude Code every day but realized I barely knew what it could actually do. Hooks? Subagents? MCP servers? I had no idea half of these existed. So I went deep over a weekend. Read the docs, tested every feature, and turned it into a game. Some things that surprised me while building this: 🔸 Claude Code can spawn parallel sub-agents to work on multiple tasks at once 🔸 Hooks let you auto-run scripts every time Claude edits a file 🔸 CLAUDE.md gives Claude persistent memory across sessions 🔸 MCP servers let you plug Claude into databases, Figma, APIs, anything 🔸 Most daily users I talked to knew maybe 2 out of these 5 15 challenges across 6 rounds. You get a spider-web radar chart and an archetype title based on your score. Built with Next.js and Framer Motion. No backend, no sign up, no email. This is a side project - not a product, not a startup. Just a weekend rabbit hole that got out of hand . What did you score? Would love to see the scorecards 👇 — Krishna
Joao Seabra

The fact that someone who uses Claude Code every day still discovered gaps in their knowledge while building this says everything about how deep the feature surface actually is.

Took the quiz and landed 87 as an Agent Architect. The radar chart is a genuinely useful mirror and the archetype breakdown is more satisfying than a single number.

One thing worth flagging: I got a clean 0 on Prompting, which felt off. Either I accidentally skipped that round, or there's something worth checking on your end, as I don't recall ever seeing that in the quiz. Also pushed back on one answer about screenshots not being a valid way to give Claude Code context, since I use visual references regularly for UI and design work. Would love to know if that's intentional. Great format overall for surfacing blind spots. Congrats on the launch! 🤖