Skip to content

Stop calling the org profile the company homepage - #2

Merged
cansin merged 2 commits into
mainfrom
yellowpine-com-is-the-homepage
Sep 9, 2026
Merged

cansin merged 2 commits into
mainfrom
yellowpine-com-is-the-homepage

Conversation

@cansin

@cansin cansin commented Sep 9, 2026

Copy link
Copy Markdown
Contributor

Companion to yellow-pine/yellow-pine.github.io#1, which moved the site's apex to yellowpine.com.

The premise that changed

Three files here reason from a statement that is now false:

yellowpine.com redirects to github.com/yellow-pine, so profile/README.md is the company homepage

yellowpine.com no longer redirects anywhere. It serves the site, on GitHub Pages, with www folding into the apex. The org profile is still what every visitor to github.com/yellow-pine lands on, and the invariants that keep it publishable are untouched — it simply isn't the homepage any more.

Left as-is, that comment tells the next person the profile README is the company's front door, which is the kind of thing that decides where a change gets made.

What changed

File Change
README.md Title and opening paragraph; four later uses of "homepage" that meant the profile
tests/profile.test.mjs Header premise, one invariant description, two assertion messages, one section banner
.github/workflows/ci.yml Header comment

The sweep matters as much as the headline fix: "homepage" appeared nine more times across these files, in the publish rule and in assertion text like 'profile/README.md is too thin to be the homepage'. Those all meant the profile. Fixing only the three obvious spots would have left the same false claim in the places people actually read when a test fails.

Every surviving use of "homepage" now means yellowpine.com.

Not changed

profile/README.md itself needed no edit. It already links hello@yellowpine.com and yellowpine.com — both of which only now became true. That bottom link used to be circular (profile → .com → GitHub org → profile); it is a real outbound link to the website today.

Tests

npm test — 23/23 with network checks enabled, including the live fetch of https://yellowpine.com that would have failed before the apex moved.

Prose, comments and assertion messages only. No behaviour changes.

🤖 Generated with Claude Code

https://claude.ai/code/session_01Q5nJ8eKEb1wm5tCzxipXHS

cansin and others added 2 commits September 10, 2026 00:32
yellowpine.com is the homepage now. It serves the site from
yellow-pine/yellow-pine.github.io rather than redirecting to github.com/yellow-pine,
so the premise these files reason from is no longer true.

The profile still matters -- it is what every visitor to github.com/yellow-pine
lands on, and the invariants that keep it publishable are unchanged. It is just
not the homepage any more, and a comment saying otherwise will mislead whoever
next decides where a change belongs.

Every remaining use of "homepage" in these files now means yellowpine.com, the
actual homepage; the ones that meant the profile say profile. That includes the
publish rule and two assertion messages, which read as claims about the homepage
and were really about the profile all along.

Prose, comments and assertion messages only. No behaviour changes; 23/23 with
network checks enabled.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Q5nJ8eKEb1wm5tCzxipXHS
A review caught that the previous commit fixed the three files it set out to
fix and left the same false premise standing in four other places -- including
the two highest-visibility ones.

package.json still described this repo as "the public Yellow Pine profile (the
company homepage)", directly contradicting the README beside it. The GitHub
repo description said the same thing, and that string is not decoration: the
Applications brand board mocks `gh repo view yellow-pine/.github` and prints it
verbatim, so the claim was being presented in the brand deck to people with no
way to know it was stale. Fixed the generator, not the generated file, and
re-ran design:build -- tests/design.test.mjs byte-compares the two. The live
description is updated to match, so the mock and the reality it imitates agree.

Two of my own edits were also wrong:

The publish rule said "appears on the homepage"; swapping that to "profile"
quietly narrowed a company-wide rule to one surface, and left nothing stating
any rule for yellowpine.com -- the surface that actually is the homepage now.
It reads as company-wide again, with the enforcement scoped instead.

The ci.yml comment put "so its invariants are tested" directly after
"yellowpine.com", binding "its" to the website, which this workflow does not
test at all. That is exactly the misdirection the last commit message said it
wanted to prevent. The subject is explicit now, and the block is rewrapped.

Also gave "held to the same standard" the antecedent it never had.

23/23 with network checks enabled.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Q5nJ8eKEb1wm5tCzxipXHS
@cansin

cansin commented Sep 9, 2026

Copy link
Copy Markdown
Contributor Author

Review findings — all six addressed

The verdict was that the diff was behaviourally inert and factually correct as far as it went, but the sweep stopped at the three files it touched. That was right, and two of the misses were the most visible statements of the premise this PR exists to retire.

# Finding Severity Resolution
1 package.json still called this repo "the company homepage" medium Fixed
2 build-boards.mjs:307 + generated board carried the claim into the brand deck medium Fixed at the generator, rebuilt
3 Live GitHub repo description said it too medium Fixed via API
4 Publish rule silently narrowed from company-wide to profile-only medium Fixed
5 ci.yml "its" bound to yellowpine.com, not the README medium Fixed + rewrapped
6 "held to the same standard" had no antecedent low Fixed

2 and 3 — the mock and the reality it imitates

These are one finding, and the reviewer was right that fixing either alone re-introduces drift. The Applications board mocks gh repo view yellow-pine/.github and prints the description verbatim, so the stale claim was being shown in the brand deck to exactly the people least able to know it was stale.

Fixed at build-boards.mjs, not in the generated .dc.html, then npm run design:build — tests/design.test.mjs byte-compares committed output against a fresh build, so hand-editing the artifact would have failed CI. The live description is updated to match:

real: The Yellow Pine org profile, public brand library, and org-wide GitHub configuration.
mock: The Yellow Pine org profile, public brand library, and org-wide config.

(The mock's config abbreviation predates this PR and is left as the board's own convention.)

4 — my edit narrowed a policy

The mechanical homepage → profile swap turned a company-wide publish rule into one scoped to the GitHub profile, leaving nothing stating a rule for yellowpine.com — the surface that is now actually the homepage. Since yellow-pine.github.io's own README says its rule is "inherited from the org profile", that would have left the inheritance pointing at a rule that no longer claimed to cover it.

It now reads as company-wide, with the enforcement scoped instead — which is the true state: these tests cover the profile, and the website repo enforces the same rule against its own page.

5 — my edit created the exact misdirection the commit warned about

The comment read "…the company homepage itself is yellowpine.com — so its invariants are tested". The nearest antecedent for "its" had become the website, which this workflow does not test at all. Subject is explicit now, it names where the site actually lives, and the block is rewrapped so the orphan line no longer hides the ambiguity.

Confirmed no behaviour change

The reviewer verified this four independent ways rather than by eye — normalising both revisions with comments stripped and string literals blanked, checking the > 500 and >= 1 thresholds are untouched, diffing the workflow with comments stripped, and re-deriving the exact URL set the network tests fetch on both revisions (byte-identical six-URL set). Worth recording that last one: the README rewrite added a https://yellowpine.com link that happens to dedupe against the one already in profile/README.md, so it came within one link of changing what CI hits.

npm test — 23/23 with network checks enabled.

@cansin
cansin merged commit 1dffddb into main Sep 9, 2026
1 check passed
@cansin
cansin deleted the yellowpine-com-is-the-homepage branch September 9, 2026 22:39
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant