ATLAS Telemetry Lab — site

The site for telemetrylab.atlas.motionapplied.com: a landing page, a page per course, and the machine-readable files that let search engines and AI assistants describe the courses accurately rather than guess.

This is deliberately a separate repository from the exercise. The exercise is a GitHub template: every learner gets a byte-for-byte copy of it. Anything that lives in the template gets copied into every learner's repository and then goes stale the moment we change it.

Layout

courses.toml              the catalogue — the source of truth for courses
index.template.html       landing page body
course.template.html      course page body
partials/nav.html         shared header
partials/footer.html      shared footer
styles.css                all styles, inlined at build
app.js                    hero animation + channel lanes, inlined at build
build/build.py            assembles pages, structured data, robots, sitemap, llms.txt
build/og.py               draws dist/og.png from the real lap
build/refresh_lap.py      re-exports public/lap.json from the lab's simulator

Build

python build/build.py

No dependencies — tomllib and zlib are standard library. Output:

Output	What it is
dist/index.html	Landing page
dist/courses/<slug>/index.html	One per course with a slug
dist/404.html	noindex, kept out of the sitemap
dist/robots.txt	Allows the AI crawlers explicitly — see below
dist/sitemap.xml	Generated from the pages actually built
dist/llms.txt	Plain-text brief for assistants (llmstxt.org)
dist/og.png	Social card, 1200×630, drawn from the real lap
dist/artifact.html	Landing page without the document wrapper, for review

Adding or opening a course

Edit courses.toml and rebuild. The catalogue, the course pages, the sitemap, the structured data and llms.txt all come from it — there is no course markup anywhere in the templates.

[[courses]]
id           = "course-02"
number       = "Course 02"
name         = "Producers and bridges"
status       = "open"            # "open", or anything else for not-yet
status_label = "Open now"
summary      = "..."
meta         = "3 modules · a weekend · free"
channel      = "Throttle"        # a channel in public/lap.json, drawn in the lane
slug         = "producers-and-bridges"   # omit and the course gets no page of its own
time         = "A weekend"
description  = "..."             # meta description + JSON-LD description

  [[courses.modules]]
  label   = "Module 01"
  title   = "..."
  summary = "..."
  graded  = "..."

status = "open" draws the course in orange with a real telemetry channel in its lane. Anything else draws it grey with a flat NO SIGNAL lane — which is what an unconnected channel looks like on an instrument, and cannot be mistaken for a launch date.

There are only two states on purpose. The words in status_label carry the difference between "in development" and "planned"; the colours do not try to, because teal against grey measures ΔE 13.0 to normal vision — below the legibility floor.

SEO and AI search

• Structured data — Organization, WebSite, ItemList, Course and BreadcrumbList, plus FAQPage generated from the page's own <details> blocks rather than a second copy of the questions, so the markup cannot claim something the page does not say.
• robots.txt allows the AI crawlers by name — GPTBot, ClaudeBot, PerplexityBot, Google-Extended and the rest. Blocking them would mean those assistants cannot cite the site, and being the answer to "how do I learn real-time telemetry" is most of the point of building it.
• llms.txt states the facts an assistant would otherwise guess at: that it is free, that no licence is needed, what the stack is, how grading works, and that there is no certificate. Everything in it is checkable against the lab repository.
• The FAQ is real content first. The eight questions are the ones engineers actually ask, answered in 40–60 words each so a passage stands on its own.

Brand

Built to the Motion Applied brand guidelines (February 2026).

MA Orange	#FA6914
MA Grey	#1F292E
Neutrals	tints of MA Grey — 94%, 88%, 80%, 60%, 40%
Extended	Cool Grey #DEE2D9 for body copy
Typeface	Montserrat, with Arial as the stated fallback

Every neutral is MA Grey or a computed tint of it, because the guidelines forbid introducing unsanctioned colours. Tint N% = N% MA Grey + (100−N)% white. The guidelines illustrate 80/60/40; the 94% and 88% steps continue that scale and exist because a dark interface needs two surfaces just above the ground.

The speed ramp on the lap and the social card is the sanctioned tint scale of MA Orange — 100%, 80%, 60%, 40% — which is monotonic in lightness, so it works as a sequential scale without inventing colours.

Typography follows the stated hierarchy: headlines and CTAs Montserrat Bold in sentence case with no closing punctuation; sub-heads with closing punctuation; body copy Montserrat Regular, which is what the guidelines specify on a dark background. Monospace is not covered by the guidelines and appears only inside <code> and the simulated terminal readout, using the system stack so no second typeface competes with Montserrat.

Contrast on the #1F292E ground: white 14.8:1, Cool Grey 11.3:1, Grey 40% 6.3:1, MA Orange 5.0:1 — all pass AA for body text. Grey 60% and 80% are 3.7:1 and 1.9:1, so they draw lines and never text. MA Grey on an MA Orange button is 5.0:1; white would have been 2.96:1 and was not used.

Not yet applied: the Motion Applied logo. The guidelines require the logo to be used with its symbol and never redrawn, and no asset was supplied — the nav and footer currently carry a text wordmark. That needs the real file before launch.

The hero lap is real

The animated lap and the social card are actual telemetry from the lab's simulator — the same code a learner runs in Module 1. It is exported once and committed as public/lap.json, so this repository builds on its own.

Refresh it only when the simulator changes:

python build/refresh_lap.py --lab ../telemetry-lab

Deploying

Netlify, from netlify.toml. Publish directory dist, build command python build/build.py, Python 3.12.

Canonical tags, og:image, the sitemap and llms.txt are all built from ORIGIN, which reads Netlify's URL environment variable. That is the site's primary address, so attaching the custom domain is enough — the next deploy picks it up and nothing here needs editing. SITE_ORIGIN overrides it for a one-off build, and the literal in build/build.py is only the fallback for building locally.