[CompleteDotTech]

Change and affected skills

Resolves #47
Resolves #48
Resolves #49
Resolves #50

footer --check --online failed because both footer logo URLs pointed at CompleteTech-LLC/.github/main/brand/, which returns 404. The footer now points at the logo masters published in this repository (assets/brand/logo.png and logo-dark.png on main, HTTP 200 with the expected hashes), and the former URLs stay in footer.legacy_logo_urls so existing READMEs still migrate safely.

• Brand data (references/brand.yaml, brand.schema.json, brand.lock.json regenerated through the CLI): the published URLs plus legacy_logo_urls, which also lists the historical company-website logo URL. All brand URLs and hosts live in brand.yaml; no brand path is hardcoded in Python.
• Footer recognition and upgrade safety (scripts/ctshowcase/readme.py, scripts/showcase.py): upgrade() replaces an old footer only for a demonstrated historical logo (a configured legacy_logo_urls entry matched on parsed src/srcset URLs, or the conventional local docs/images/logo.svg) together with the strong footer signature. Alt text alone is never provenance. Images are classified from whole parsed attributes (exact names, first occurrence wins, including valueless first attributes), so data-alt, title text, spoofed hosts, userinfo and unexpected ports never count. Authored partner and foreign images, authored anchors and the generated banner's link wrapper are preserved; a change that would lose one is flagged by the preservation guard and the CLI refuses the write. readme --bootstrap --write refuses to regenerate a marked footer that holds an authored image in EVERY bootstrap state (none, pristine and edited: it fails with readme-shrink and leaves the README byte-unchanged), and the G/H unmarked website-logo and docs/images/logo.svg footers now migrate. The two existing legacy-footer tests are unchanged.
• CI (.github/workflows/footer-assets.yml, job footer-online): pull requests run the OFFLINE step that the configured logo files exist in the checkout under assets/brand/. The real footer --check README.md --online HEAD runs on push to main (footer-related paths), on the weekly schedule and on manual dispatch, with contents: read only. This matches issue footer --check --online fails: both footer logo URLs return 404 #47 ("a scheduled check or CI step fails when they do not") and E2.4.3 Add a CI check with an offline-friendly mock #50 ("offline-friendly mock", "real HEAD on a schedule", mocked unit tests for the gate logic); the online step has not yet run in hosted CI and is first exercised after merge (a manual dispatch will confirm).
• Docs: references/decisions.md keeps decision 18's quoted markup verbatim and explains the configured URLs in a separate, clearly labelled skill-maintenance note; references/readme-variants.md (the upgrade row now says a logo qualifies by its parsed src/srcset or the local docs/images/logo.svg, never by alt alone), references/art-direction.md, the changelog, examples and assets/brand/README.md describe the published URLs and the actual CI behaviour. No brand mark was created or redrawn.
• Tests: tests/test_footer_public_assets.py (offline mock of the published assets), tests/test_footer.py, tests/test_footer_fenced_example_migration.py, tests/test_footer_upgrade_preservation.py and attribute-token tests cover the partner, foreign, spoof, valueless-attribute, banner-wrapper, bootstrap and G/H cases; synthetic fixtures only.

Prior work

Searched this repository's PRs and issues for the footer logo URL problem (search repeated before this refresh):

• PR Repair footer logo URLs and safely migrate existing footers #19 (merged, dd720f9) "Repair footer logo URLs and safely migrate existing footers" made this same repair: published footer.logo_*_url values, legacy_logo_urls, safe migration of existing footers, and tests.
• PR Add opt-in T3 showcase support and Pick Your AI identity #20 (merged, f9ec72a) "Add opt-in T3 showcase support and Pick Your AI identity" merged afterwards and restored the old CompleteTech-LLC/.github/main/brand/ values in references/brand.yaml, so main is back to the 404 URLs. footer --check --online fails: both footer logo URLs return 404 #47 is a regression of Repair footer logo URLs and safely migrate existing footers #19's repair, not a new design.
• Issue footer --check --online fails: both footer logo URLs return 404 #47 and sub-issues E2.4.1 Find where the logos actually live #48-E2.4.3 Add a CI check with an offline-friendly mock #50 are the only issues covering this; no other open PR touches the footer URLs. This PR reuses Repair footer logo URLs and safely migrate existing footers #19's approach on top of current main, recovers the legacy local-logo footer recognition, and adds the footer-online CI guard (sub-issue E2.4.3 Add a CI check with an offline-friendly mock #50) so a future merge cannot silently revert it.

Provenance

Commits dfc93ed03acf4d23c187c33ada1ebfbf4fe70dbc and 5a3d78d8de9e0eac7f59d3e70b8ff5a5873a0373 are inherited, already-published history that predates the current remediation. They carry generic Co-Authored-By: Claude trailers; no more specific model is claimed for them because no native evidence is available.

Three already-published merge commits in this branch have no Co-Authored-By trailer. Commit metadata is immutable once published and the user's instruction forbids force-push, so these are recorded as provenance exceptions, not defects fixed by changing source:

Merge commit	What it merges	Made by (native evidence)
366c735c994f361c13a6259e36ef1c67d47297eb	hk/47-r3 decisions.md skill note (Haiku commit d5c63ca, Haiku trailer)	Sonnet owner integration fix:#47:r3, native agent a755d4e54043aaf95, runtime model claude-sonnet-5-5
76ee8b61fe4323f8da22e8eff470c9c5d37694ac	hk/47-r3 README footer replacement fix (Haiku commit 740303d, Haiku trailer)	same Sonnet owner integration fix:#47:r3, a755d4e54043aaf95, claude-sonnet-5-5
7fc3743b951173e02fc447a5d32f5fa409d2654d	849f67e (Haiku commit, Haiku trailer)	Sonnet owner integration fix:#47:r2, native agent a5f0db6ef29599f82, runtime model claude-sonnet-5-5

Nothing is rewritten, no trailer is claimed that does not exist, and no model beyond the native evidence is asserted. From here on, every commit and merge commit carries the verified trailer of the role that made it before it is published (Haiku 5.5 for file workers, Sonnet 5.5 for owner commits and merges). This is a provenance record only; it is not an approval of the code.

Additional integration merges without a trailer (the model trailer on merge commits is a convention of the orchestration prompt, not a repository rule; recorded, not rewritten, no attribution invented):

Merge commit	Adopts	Made by (native evidence)
cad87ddf1662e5f3c5088217255fce891ec1922f	Haiku commit c017dd14e3e318e6ac700b3653f8eaeb17e05417 (decisions.md wording)	Sonnet owner integration fix:#47:r5, native agent a7a484513792f76f1, claude-sonnet-5-5
512f13e4b707c7c6e6b2a14fdde56362bd9dad52	Haiku commit 2b0e0b0723b96429a39ff750e20aae66003304b8 (art-direction.md)	same Sonnet owner integration a7a484513792f76f1, claude-sonnet-5-5

Further integration merges without a trailer (convention of the orchestration prompt, not a repository rule; recorded, not rewritten): 9a5ae0d (adopts Haiku d4d4c0b, claude-haiku-5-5) and the e58465e/97b775d line, made by the Sonnet owner integrations fix:#47:r0 (native agent a6f7eade5) and fix:#47:r1 (native agent afefb0a7e), both claude-sonnet-5-5; the Haiku source commits carry Haiku 5.5 trailers.

Later history on this branch, recorded as it happened (nothing rewritten): commit 384f443 briefly changed the CI workflow to run the online HEAD on pull requests, which contradicts the real acceptance of #47/#50; 0f1eec8 restored the workflow and its original test byte-for-byte (verified against bec6391), and 76f0108 removed a duplicate test helper flagged by Ruff F811; d17909e extended the bootstrap refusal to the none and pristine states; 7a72de1 corrected the readme-variants logo wording, pinned the CI policy in a test and normalized one new test file to LF. The merge commits after the table above follow the same convention (integration merges by the Sonnet owner, Haiku source commits carry Haiku 5.5 trailers); this section is a provenance record, not an approval.

Evidence (exact head 7a72de1c706a6f5ee815aa28d7589c90620ad13b, base main 85f12d8198f77a1bea150a9b0afff0125af9e990 integrated by a normal merge)

• Isolated VM (Ubuntu 24.04, run by the outer coordinator; receipt vm-7a72de1c706a6f5ee815aa28d7589c90620ad13b-result.json): package tests, package validator, full pytest and quality validator passed; pytest 1198 passed, 11 skipped, 37 subtests. The skips are the pre-existing absent external fixtures; the Windows-only symlink cases run in Linux; no jsonschema or changed-feature skip.
• Footer probes on the same VM: seven footer probe groups, the r7 same-host-photo and anchor controls, bootstrap and G/H cases, all three bootstrap states (none, pristine, edited: each refuses the write and keeps the authored partner bytes), and fresh public logos (HTTP 200, downloaded sha256 equals the committed masters: logo.png 243500 bytes a5c3c4ef...1621, logo-dark.png 210785 bytes 0dc00c0e...7ec) all pass. A deliberately missing asset fails with fail:offline (negative control).
• Hosted CI on this head: validate-package (ubuntu, windows, macos), pytest (ubuntu, windows, macos), validate and footer-online all completed with success (15 check runs, push and pull-request matrices). One macOS validate-package job was left by GitHub in an inconsistent state (status in_progress with conclusion success while its workflow run was completed); it was rerun as a targeted job rerun, the original record is preserved, and the rerun completed with success.
• Local Windows run at this head (supplemental, guarded single full run): 1196 passed, 13 skipped, 37 subtests (the two extra skips are the Windows symlink-privilege skips that run in the VM). Earlier local Windows runs on other heads failed at fixture git commit setup in the local signing environment; those logs are preserved and are not evidence for this head.
• Independent reviews of this exact head and base (native provenance verified): two Opus reviewers (claude-opus-5-5) and Codex gpt-6.1-sol at max effort each returned an explicit approve with zero findings. An earlier Sol review of this head returned incomplete (it had no network and the live HTTP receipts were not in its context); the receipts are now supplied inline and the incomplete record is kept. Earlier reviews of older heads (including a lifecycle-only request for explicit closing keywords, resolved by the keywords above) do not carry over.
• Not run: mmdc/Mermaid rendering (tool unavailable); private deny-list check (no private fixtures); pyright (no config).

Limitations

• The real online HEAD step has not run in hosted CI yet because pull requests run only the offline file-existence step; its first run is on push to main, the weekly schedule or a manual dispatch (to be triggered after merge).
• Non-blocking reviewer suggestions tracked for later cleanup: an unreachable second authored-footer refusal in the bootstrap edited branch, broader path filters for footer-assets.yml (scripts/ctshowcase/readme.py, references/brand.lock.json), a clearer error if a logo URL is ever off main, and narrowing the historical bare local logo forms. A follow-up issue will cover readme --mode header, which still regenerates a marked footer without the authored-content refusal (pre-existing, outside this change).

Boundaries

Activation keys, consent, approval gates and specialist entry points are unchanged. Branding decisions are not altered beyond repairing the unreachable URLs (the quoted decision text is untouched). No registry release, visibility change or live external action is implied. Shared validators (scripts/validate_package.py, tests/test_package_contract.py) are byte-identical to main.

Family consistency

Single-repository change; no sibling skill PRs are affected.

🤖 Generated with Claude Code

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
🔒 Security Review	⚠️ Failed 2026-10-31T19:07:28.719453Z We couldn’t complete this review. Comment @codex security review to try again.	5a3d78d	PR opened

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[chatgpt-codex-connector]

The account paying for this security review has reached its Codex usage limits. The payer can check the Codex usage dashboard. For personal accounts, using credits requires enabling “Use credits for security reviews” in Code review settings. If you do not manage the paying account, contact this repository's admins.