TokenQuest

Somewhere in your ~/.claude directory, a hero has been questing for weeks. Every prompt you typed was a spell. Every 3 AM debugging session, a night march. Every compaction survived, a memory reforged. TokenQuest reads that history — entirely on your machine — and hands you the character sheet you already earned: level, eight attributes, a class, a chronotype race, gear, a spellbook, and a place in the Halls of Heroes.

No account. No email. Nothing to watch — it plays itself while you work. The first thing it does is replay your history: “you're already Level 34.”

The short version of how this works

Do it yourself (the four commands)

curl -fsSL https://questful.online/get/tq.py -o tq.py
python3 tq.py --dry-run                          # see what a push would send (sends nothing)
python3 tq.py --register --handle YourHeroName   # claim your hero (prints your recovery passphrase ONCE)
python3 tq.py --play                             # your game, at http://127.0.0.1:8123 — pushes every 5 min while it runs
# on Windows: replace python3 with py -3

The mirror, plainly (--mirror on — optional, off by default): normally your game exists only on your machine. The mirror asks us to host a copy of your living panel at a share link, so you — or anyone you hand the link — can watch your hero from any browser, phone included, even while your machine is off. It also acts as an off-site backup of your history. The trade: each push then uploads your panel's display feed too, scrubbed first (details below) — though still carrying your timezone and your first-seen date: to anyone holding the link it is your daily activity calendar. The link ends in a rotatable watch key — a view key, not a credential: --mirror rotate invalidates an escaped link, --mirror off deletes everything we stored.

Keeping it running: --play is a plain foreground process — it stops when your terminal closes or your SSH session ends. Run it under tmux (or nohup python3 tq.py --play &) to keep the local dashboard up; the printed cron line keeps your hero fresh on the board either way. And if what you really want is “see my game from anywhere, no process needed” — that is exactly what the mirror is for. Runs happily on a VPS — if you use Claude Code over SSH, run it there; VS Code forwards the port automatically. One caveat for shared machines: the local server trusts the local machine — any account on the same box can read your unscrubbed feed, project names included, so run it where you're the only user, or start it with --local-key yourkey: then every request needs ?token=yourkey (the printed URL includes it). --reclaim brings your hero to a new machine.

Windows & locked-down networks: on Windows, python3 opens the Microsoft Store — use py -3 (or python); tq.py's own printouts use the right command for your OS. If the download fails oddly, or --register/--push dies with a certificate error on a corporate network, your proxy is likely re-signing TLS — tq.py will say so, and the one-line fix is NO_PROXY=questful.online (or download tq.py in your browser instead of curl). Some corporate filters also block this domain as “newly registered” — that needs an IT allow-list; nothing on your end can fix it.

One hero per project — and the party camp: by default all your Claude work feeds one hero. If you'd rather split it, --profile work --projects "client*" makes a named sub-hero fed by only the projects that match (it can register on the board as its own hero, under any name you like — tip: not the client's). Once you have profiles or camp members, --play opens on the party camp: all your heroes around one campfire, and the fire itself is the party's number — it grows with your heroes' summed adventures and never shrinks. Each hero's own world is one click away. Play on several machines? Two stories, pick per machine: --link folds another machine into this hero — one character, the sum of all your machines, on the board and on every machine's own panel (mint a pairing code with --link-code on a set-up machine, run --link <code> on the new one; your passphrase never travels). Or, for machines — or people — you want visible as separate heroes, register a hero on each, then --camp add tokens/YourOtherHero seats the others at this box's fire — your whole party, one camp. The camp page itself talks only to your own 127.0.0.1 server; for camp members, your agent reads their public Hall standings from the board (the same numbers /halls shows), only while the page is open. Curate with --camp add / remove / list; it's a plain JSON file (~/.tokenquest/camp.json) if you'd rather edit by hand.

What leaves your machine — the whole payload

This is a real --dry-run output, in full. Fourteen aggregate measurements, a motto, an avatar. No project names, no file paths, no prompt text, no daily series, no histograms. The server derives your stats from these — the formulas live in one place, so the agent on your disk never needs updating.

{
  "measurements": {
    "user_prompts": 821,          "total_active_min": 7187,
    "avg_session_min": 37,        "active_days": 24,
    "corpus_days": 25,            "tokens_total": 3428299977,
    "tokens_io": 14294654,        "user_words": 30004,
    "user_chars_typed": 171303,   "compactions": 0,
    "agent_launches": 8,          "workflow_uses": 0,
    "peak_day_io_tokens": 1481071, "cost_io_usd": 344.97
  },
  "motto": "120h adventured · 14.3M tokens slain",
  "avatar": "⚔"
}

The opt-in mirror sends more (your panel's display feed) — scrubbed first: project names, your --server machine label, the weekday-hour matrix and unrecognized MCP tool names never leave your box (your timezone and first-seen date stay in — the panel needs them to draw your days). --dry-run summarizes that payload too when the mirror is on: its size, what was scrubbed, what's kept. Off by default.

Opt-in machine linking (--link) sends the same class of extra: each linked machine's push uploads its per-day tally sheet — exactly the mirror's scrubbed shape, never transcripts, prompts, code, or project names — and downloads its siblings', so every machine (and the board, and the mirror if on) shows the one summed hero. Sheets stay private to your hero's own machines — device-token read, never published. --dry-run notes the exchange when linked; --devices list / remove manages machines, removal is instant.

Fair warning about the paste-a-prompt method

Pasting instructions into your agent is exactly the pattern security guidance tells you to distrust — so distrust it: the message below tells your Claude to download tq.py, audit it, and explain it to you before anything runs, and to stop if anything looks off. The top of tq.py carries THE TRUST MAP — an exhaustive list of every network request the script can make — written to be checked against the code, not taken on faith. Auditing us first isn't paranoia here; it's the intended on-ramp. If you'd rather stay hands-on, the four commands above do the same thing.