Robert Blust's profile page and its talks. Self-contained, no bundler, no external assets. Working conventions and the traps that break this site silently are in AGENTS.md.
/ profile page — thesis, links to /talks/
/talks/ talks index — the descriptions live here
/talks/mental-model/ deck: The Mental Model
/talks/essential-complexity/ deck: Essential Complexity
/talks/deciding-well/ deck: Building fast is solved. Deciding well is not.
/talks/deciding-well/cost/ what a team would have cost — computed from the talk's stats.json and cost.json
/surfaces/ every place the model is published and how each is made — reads the same model.json as /model/
/processes/ each process and the seats it names — reads the same model.json as /model/
/timeline/ the experiences as a ledger — reads the same model.json as /model/
A talk is named in one place. talks/index.html is the only page that names a talk; the profile page links to /talks/ and lists nothing. Adding or editing a talk is therefore one edit, not two — see AGENTS.md for why the second one is gone.
Each deck is bilingual (English content, German via data-de), links ../../tokens.css, ../../deck.css, ../../deck.js and the root fonts/ rather than standing alone, and works from file:// as well as a local server.
npm install && npx playwright install chromium
npm run serve # python3 -m http.server 8000
npm run verify # Playwright DOM assertions against every page, plus the sitemap
npm run og # regenerate the 1200×630 og:image share cards
npm run og:check # do those cards still show the pages they were rendered from?
npm run test:og # unit tests for the card recipe the check compares
npm run pdf # regenerate every deck's PDF fallbacks
npm run model # fetch the pinned model and write model.json — needs the network
npm run model:check # is model.json still what that commit parses to?
npm run pages # render model.json into every derived page region
npm run pages:check # do those pages still match model.json?
npm run test:build # unit tests for the renderers
npm run sitemap # date each sitemap URL from its page's last commit — run before committing a page
npm run sitemap:check # are those dates still what git says?
./tts/generate.py --dry-run # narration: what would be billed, and for which slides
./tts/generate.py # narration: generate what changed, every deckRun npm run verify after any change under index.html, talks/, or verify/. Run npm run og and npm run pdf after a visual change to either deck or to index.html / talks/index.html — the share cards and PDFs are rendered, committed files, not generated on demand. npm run og:check says when a card has fallen behind its page; CI runs it on every push, so forgetting is caught rather than shipped. Run npm run model then npm run pages after moving the pin in source.json; model.json and every page built from it are committed files, not generated on demand.
Narration is generated from the speaker notes themselves and cached on a content hash, so editing one note regenerates one clip. It needs ELEVENLABS_API_KEY, which lives in ~/.zshrc and is therefore invisible to a non-interactive shell — AGENTS.md has the one-liner that reaches it, and the rule about never printing it.
CC BY 4.0 for everything written here: the pages, the talks and their narration, the share cards, the picture, and the scripts that build and check them. Use it, quote it, build on it; credit it. Four kinds of file are not written here and keep their own licenses, each in a text beside it: the faces under fonts/, d3.v7.min.js, mermaid.min.js and the GitHub mark inlined in chat.js come from @robertblust/design, whose NOTICE names each one. The license grants no right to a name or a mark: the rb mark and the name stay mine.