seriesoftubes/CLAUDE.md
Frederic Jung 308c3f7045
Performance benchmark on a big network, posted on every PR
scripts/perf-bench.js times routing, a game day, an action, panels, the
NOC dashboard and the map on tests/fixtures/big-network.json (104 POPs,
950 PNIs, 832 caches), base and head in alternating turns, and says
whether routing results changed. perf.yml runs it on every PR against
the base; perf-comment.yml posts the table as one updated comment.
New performance skill, CLAUDE.md rule, PR template and docs.
2026-10-26 06:39:23 +01:00

4.3 KiB

Series of Tubes: notes for Claude

A free cartoon browser game about how the internet works, live at https://seriesoftub.es. The whole game is one static page: site/index.html (HTML, CSS and JS in one file, no framework, no build step for development).

Read these before larger changes:

  • docs/ARCHITECTURE.md: how the code is laid out and how a day of simulation runs.
  • .claude/skills/: step-by-step procedures (release, translations, tests).

Commands

npm ci                       # once: installs jsdom for the tests
npm test                     # node:test + jsdom, all suites, must stay green
node scripts/i18n-check.js   # must report "0 missing, 0 unused, 0 broken"
npm run build                # builds _site/ (stamps version, bakes release notes)
npm run serve                # build and serve locally
npm run perf -- --base /tmp/base.html   # big-network benchmark vs another version (see `performance`)

To play during development, open site/index.html directly or run npm run serve.

Hard rules

  • Git: work on main, no branches. Commits are authored and committed as Frederic Jung <frederic@jc-ix.net>. Never add Co-Authored-By, Claude-Session or any other attribution trailers.
  • Releases only when Frederic says so. Never bump the version on your own: push finished work with its CHANGELOG bullets under ## [Unreleased] and say it is ready. When he asks for a release, follow the release skill: small fixes get a patch (x.y.Z), features a minor, at most one release per hour. Docs, skills, CI, tests and refactors never need one.
  • Release notes in the app are short: only the first sentence of each CHANGELOG bullet (and at most five bullets) shows in "What's new", with a link to the full GitHub release. Write that first sentence so it stands on its own.
  • Names are puns. No real company names for networks, vendors, cables, banks, data centers or exchanges (Cogentle, Goggle, NetChill, Junipurr, Arrrhipster…). Real conference names (NANOG, RIPE, DENOG) are fine. Avoid lazy "-ish" names: find a real pun, or keep the plain word.
  • Easter eggs stay secret. Never mention hidden jokes or easter eggs (for example whatever elephant() does) in commit messages, release notes or the CHANGELOG.
  • Every player-facing string is translated. Wrap it in tr(), trn() or N_() and add the Italian to site/i18n/it.js in the same change (see add-strings).
  • Phones first. About 70% of players are on phones. Every new panel, button and modal must work at 360 px wide, in portrait and landscape, using the phone tab bar and bottom sheets. Check isPhone() layouts, not just desktop.
  • Saves are sacred. Old saves must keep loading. Any new state field gets a default in newGame() and a check/repair in sanitize(); renamed fields are migrated there.
  • Performance is always demonstrated. Every change to site/ is measured on the big-network fixture against main (npm run perf -- --base /tmp/base.html, see the performance skill), and the numbers go into the PR and into your summary. Every PR also gets them as a comment from CI. Anything slower needs a reason or a fix.
  • Tests come with features. Add or extend a test in tests/ for every feature or fix. Tests are seeded, so they must not depend on luck.
  • CHANGELOG style: plain English, written for players, one bullet per change, issue numbers in parentheses like (#37). No internal names, no jokes explained.

Style of the code

  • Dense, single-file JavaScript. Match the surrounding style: short helper names, arrow functions, template-literal HTML, no new dependencies.
  • UI is rebuilt from state: change S, then call after() (recompute, check goals, re-render, save). Buttons are btn(label, actName, args, cls, disabled), which dispatch to ACT[actName](args) through data-act/data-args.
  • Never trust saved or imported state: anything read from S that came from a save must already have been validated in sanitize().
  • Keep comments short and about why.

Tone

Friendly, nerdy and accurate underneath the jokes. Mechanics should follow how networks really work (95th-percentile billing, peering policies, LAGs, DWDM reach, route leaks, RPKI) and then exaggerate for fun. When unsure how something works in real life, look it up before modelling it.