fix(validate): the scan missed four kinds of file, and vouched for prose

[pitimon]

Summary

Closes #110 items 2, 6 and 1. Stacked on #111 — review that first, or read this diff two-dot against fix/110-comment-and-pinned-waivers.

Item 2 — the walker saw two directories and one extension set

Added dashboard/*.html (the served shell — a <link rel=preconnect>, font CDN or <script src> here is a browser request), .css (skipped by extension while the walker was already standing in that directory), i18n .json, dashboard/vite.config.js, and scripts/.

That surfaced eight things nothing had ever mentioned. Each was read in source before being classified — declaring without looking would just be a waiver nobody reviewed:

Found	What it actually is	Declared as
scripts/build-pricing-seed.cjs:10	really fetches the LiteLLM price list	request_from
src/lib/pricing/seed-snapshot.json:1	the seed recording its source URL	data_from pin
dashboard/vite.config.js:1188	dev server proxying /proxy/ipcheck	request_from
dashboard/index.html ×5	footer links + JSON-LD sameAs to the upstream repo	link_from pins
www.tokentracker.cc	JSON-LD url/logo/breadcrumbs — search-engine metadata, not fetched	new entry, data_from
www.npmjs.com	JSON-LD downloadUrl. Distinct from registry.npmjs.org, which npx really contacts	new entry, data_from
schema.org	the JSON-LD @context identifier, same class as the www.w3.org namespace	ignored_hosts
example.invalid	the reserved TLD that never resolves, used as a dead endpoint in acceptance scripts	ignored_hosts

Item 6 — the inventory's prose was unchecked

user_data: true with readme: false passed. So did a wrong from, and purpose text describing something the code does not do. The fields a reader trusts most were the ones nothing verified.

Types and required values are now checked, and from is checked against where the host is actually reached — a host only ever requested from src/ cannot honestly be browser.

Proved non-vacuous against five deliberately-wrong entries before committing, with a valid control that stays clean.

Item 1 — unresolvable authorities were dropped

The issue's three examples — bracketed IPv6, a Cyrillic lookalike, a single-label intranet host — all resolve now. They predate deferring to new URL(). What is left is interpolation that pins nothing, which is ordinary code, which is exactly why the flag-flip version had to be reverted inside #109.

So each is declared with a why, matched on file + authority text, and a new one fails. 15 declared across 8 files: the local bind and Host-header parsing, git-remote normalisation into project_ref, the user's own HTTPS_PROXY, dev-server proxying, and this scanner's own probe strings. A declaration that no longer matches the code is reported as stale, so the list cannot outlive what it excuses.

Test plan

• ci:local exit 0 — 903 root tests (+12 over fix(validate): trailing comments read as requests; waivers covered whole files #111), 256 dashboard
• validate:outbound green: 26 declared hosts
• 19 new tests: the shell, CSS url(), scripts/, the dev config; five metadata violations plus a valid control; a new unresolvable authority fails, a declared one passes, a stale declaration is reported
• Two existing tests updated where the change inverts their expectation — an unpinnable authority is now reported as unresolvable rather than silently dropped. Both still assert no host is invented, which was their original point.
• checkOutbound (46) and collectHosts (21) back under the 50-line rule via scanLine (39), checkUnresolved (30), checkInventoryMetadata (29). File 640 lines.

Remaining in #110

Item 5 (URL composition — following expressions rather than scanning literals) and item 7 (avatar body cap, which is PR #112).

[pitimon]

Superseded by #114. This PR was auto-closed when its base branch fix/110-comment-and-pinned-waivers was deleted on merging #111, and GitHub will not reopen or retarget a PR whose base is gone. #114 is the same commit cherry-picked onto main, so it also gets real CI — which this one never had, since the workflow only triggers on pull_request: branches: [main].