Make the crawlable index actually reachable, and fix deep-link photo URLs

[ledwards]

Two fixes on the way to making root.vc's terminal legible to crawlers without a static mirror.

The crawlable index was invisible to the crawlers it was written for

The homepage block listing every portfolio company and team member was wrapped in <noscript>. That loses it twice over:

• Text-extraction pipelines (Readability-style, trafilatura, most naive HTML-to-text passes) routinely strip <noscript> as non-content. An LLM crawler running one of those sees an empty page.
• Googlebot indexes the rendered DOM, and <noscript> content is absent from it once JS runs.

So the one artifact meant to make the terminal readable was skipped by most of its intended audience.

It's now an offscreen div using the .visually-hidden class already in css/styles.css — whose own comment notes that display:none gets discounted by search engines. The text sits in both the raw HTML and the rendered DOM, and sighted visitors still never see it.

Photo fallback URLs broke on deep-link arrivals

js/ascii-art.js built the fallback text by concatenating onto document.location.href, so any fragment ended up in the middle of the path. Landing on /#whois-avidan rendered:

[ Photo: https://root.vc/#whois-avidanimages/avidan.png ]

Resolving against document.baseURI drops the fragment the way URL resolution is meant to. This matters more now that deep links are becoming the main way to reach specific content.

Verification

• npm test — 313/313 passing, including a new test that locks in the no-<noscript>/no-display:none decision so it doesn't get "tidied" back
• Built dist/ and served it in a browser: #text-version is in the rendered DOM, clips to 1×1, adds no scroll overflow, and a plain text extraction of the page returns all 6318 characters
• Screenshotted /#whois-avidan before and after — terminal visually identical, photo URL now correct

🤖 Generated with Claude Code