Gamify your Claude Code sessions — track token efficiency, earn achievements, level up your prompting.
npx claudepluginhub josheche/tokengolfGamify your Claude Code sessions — track token efficiency, earn achievements, level up your prompting.
Every token counts. Every session is scored.
tokengolf.dev · npm · GitHub
Turn Claude Code token efficiency into a game. Every session is automatically tracked and scored against a dynamic par budget. Work normally. On /exit — a scorecard, achievements, and a score that measures how well you prompt.
Better prompting → fewer tokens → higher score. 60+ achievements. 9 hooks. 4 character classes. Zero config beyond install.
██ 🏆 SESSION COMPLETE
██ ──────────────────────────────────────────────────
██ $0.21 /$3.00 par 14% ⚔️ Sonnet 🥇 Gold
██ 🌟 LEGENDARY
██ ──────────────────────────────────────────────────
██ 🎯 Sniper 🥈 Silver 🔥 No Rest ✅ Clean Run 🧰 Toolbox 🤫 Silent Run
Claude Code Plugin (recommended — auto-updates)
claude plugin marketplace add josheche/tokengolf
claude plugin install tokengolf@tokengolf
npm (alternative — requires hook setup)
npm install -g tokengolf
tokengolf install # patches ~/.claude/settings.json
Homebrew
brew tap josheche/tokengolf
brew install tokengolf
tokengolf install
curl
curl -fsSL https://raw.githubusercontent.com/josheche/tokengolf/main/install.sh | bash
That's it. Open Claude Code, work normally, /exit — scorecard appears automatically.
npm users get auto-sync — hooks update automatically on version change.
tokengolf scorecard # last run scorecard
tokengolf stats # career dashboard
tokengolf config # view all config values
tokengolf config emotions emoji # set emotion mode (off, emoji, ascii)
tokengolf config par # view/set par rates per model
tokengolf config floor # view/set par floors per model
tokengolf demo # show all UI states (hud, scorecard, stats)
tokengolf install # patch ~/.claude/settings.json with hooks (npm only)
Every Claude Code session is automatically tracked — no setup, no pre-configuration. TokenGolf measures your efficiency against a par budget that scales with your session:
par = max(rate × sqrt(prompts), model_floor)
efficiency = actual_cost / par
Par grows with your session, but sublinearly — early prompts give you room to explore, while pressure builds as the session goes on. Efficient prompts beat par; wasteful ones fall behind. Spend more than par and the run busts — logged as a death with red accents and death achievements.
| Model | Par Rate | Floor |
|---|---|---|
| 🏹 Haiku | $0.15 | $0.10 |
| ⚔️ Sonnet | $1.50 | $0.75 |
| ⚜️ Paladin | $4.50 | $2.00 |
| 🧙 Opus | $8.00 | $3.00 |
The floor prevents 1-prompt sessions from being instant BUST. The sqrt scaling means a 4-prompt Sonnet session has par $3.00, while a 16-prompt session has par $6.00 (not $24). Long sessions must be increasingly efficient to stay under par.
Rates and floors are configurable per model:
tokengolf config par # list all rates
tokengolf config par sonnet 1.5 # set Sonnet par rate to $1.50/prompt
tokengolf config par reset # restore defaults
tokengolf config floor opus 20.0 # set Opus floor to $20.00
tokengolf config floor reset # restore defaults
| Class | Model | Difficulty | Feel |
|---|---|---|---|
| 🏹 Rogue | Haiku | Nightmare | Glass cannon. Prompt precisely or die. |
| ⚔️ Fighter | Sonnet | Standard | Balanced. The default run. |
| ⚜️ Paladin | Opus (plan mode) | Tactical | Strategic planner. Opus plans, Sonnet executes. |
| 🧙 Warlock | Opus | Casual | Powerful but expensive. |
RuFlo Marketplace: Claude Code native agents, swarms, workers, and MCP tools for continuous software engineering
No description available.
Code intelligence powered by a knowledge graph — execution flows, blast radius, and semantic search