About tokenmogged

Mid prompts. Mid code. Mid life.

Two coders. One topic. 45 minutes. Your Claude vs theirs. Better implementation wins, and the ELO ladder remembers.

How a match runs

1. Sign in (GitHub, Google, or email, whatever).
2. Run curl https://tokenmogged.com/install | sh. The CLI lands on your PATH and wires the Claude Code hooks. Two follow-ups on first run:
   • tokenmogged login: opens a browser page, you copy a short code back into the CLI, you're bound.
   • tokenmogged setup: drops the hook config into ~/.claude/settings.json. Safe; doesn't touch anything else there.
3. Hit Play. Matchmaker pairs you against the nearest ELO. Or send a challenge to a friend who needs humbling. Unranked, just bragging rights.
4. Both ready up. Topic reveals. The CLI auto-launches claude in a fresh scratch dir under ~/.tokenmogged/matches/. Build the thing. Opus, Sonnet, Haiku. Your call. Switch freely.
5. Match ends on first compaction, 1M tokens, or 45 minutes, whichever hits first. Or type /exit when you're done shipping.
6. Match ends. ELO moves. You land on a review screen with both code submissions side-by-side.

Friendships

Search a username, send a request, accept it. Then challenge each other to unranked matches. Same engine, same review screen, but ELO stays out of it. For when you want to mogg without consequences.

The CLI

You bring the Anthropic subscription. We bring matchmaking, the ladder, and the cope. The CLI is MIT-licensed and public. Read it before you install. Or don't. Skill issue either way.

I'm busy. If you want to run this, hit me up.