WebMCP Dino

checking for WebMCP…

idle HI 00000   00000
Space / ↑ jump · ↓ duck
Let Claude Code play this game
  1. Add the WebMCP bridge to Claude Code
    claude mcp add webmcp -- npx -y @smarchone/webmcp-bridge
  2. Ask Claude to play
    claude "Open https://smarchone.github.io/webmcp-apps/dino/ with the webmcp tools and play one game"

Run both in the same folder. Requires Node 22+ and Chrome.

Also works with other local agents that support MCP, like Codex, Gemini CLI, Cursor and VS Code. Add the bridge to their MCP config.

Advanced

For better play, replace step 2 with either of these.

  1. A detailed prompt
    claude 'Open https://smarchone.github.io/webmcp-apps/dino/ with the webmcp tools. Call get_game_rules first and follow its physics and strategy. Play 3 games. After each one, call get_history, say what ended the game and adjust your trigger gaps before the next.'
  2. The /play-dino command, then run /play-dino 3 in Claude Code
    mkdir -p .claude/commands && curl -fsSL https://smarchone.github.io/webmcp-apps/dino/play-dino.md -o .claude/commands/play-dino.md

Stats

Speed
-
Games played
0
Agent best
-

In agent mode the game freezes when an obstacle comes into range. The agent plans when to jump or duck, and the game carries out the plan in real time.

Agent decisions