test: refresh blog visual regression baselines (macOS + Linux)

[pftg]

Summary

Refreshes the 3 stale blog visual regression baselines on both macOS and Linux that have been failing on master with identical `difference_level: 0.062389564043209875`. Verified on both runners: all tests pass against new baselines.

Root cause

Baselines last captured 2026-04-14. Since then 5 intentional blog UI changes shipped without updating baselines:

Date	Commit	Change
2026-04-30	`a7c5cd3e`	fix(blog): index card tag taxonomy + ICP hero copy
2026-05-02	`566044d1`	Compact reading sections + 6 post engagement improvements (#311)
2026-05-03	`b4d6d1e9`	feat(blog): centered single-column typography (post + index)
2026-05-04	`f1aeb21b`	Use explicit `.post-description` class for blog listing description (#320)
2026-05-05	`059157c4`	fix(blog): show descriptions on mobile + fix overflow at narrow viewports

Each shipped while the visual regression tests were already failing — the gate fired but nobody updated baselines. The 0.062 difference is the cumulative drift, not a single regression.

Investigation path

1. Reproduced on master — same exact `difference_level` as on PRs seo: consolidate fractional CTO pages (fix position-75 cannibalization) #326, seo: consolidate Rails perf posts into 2026 canonical (hub-and-spoke) #327, seo: new /services/react-development/ page (Block 3) #328. Confirmed pre-existing.
2. Diff image (`index.diff.png`): yellow boxes around blog post cards (dynamic content), not on layout/CSS. Content drift, not layout drift.
3. Checked baseline file dates vs blog template change history. Baselines from April 14, 5 intentional changes shipped after.
4. Examined gem source (`snap_diff-capybara`): baseline pulled via `git show HEAD:`, not from working tree. Working-tree updates do nothing until committed.
5. Captured new baselines via `FORCE_SCREENSHOT_UPDATE=true`, committed, re-ran tests on the new HEAD — all pass.

What changed

macOS baselines (commit `6321fe9e`):

• `test/fixtures/screenshots/macos/desktop/blog/index.png`
• `test/fixtures/screenshots/macos/mobile/blog/index.png`
• `test/fixtures/screenshots/macos/mobile/blog/post.png`

Linux baselines (commit `918f8668`):

• `test/fixtures/screenshots/linux/desktop/blog/index.png`
• `test/fixtures/screenshots/linux/mobile/blog/index.png`
• `test/fixtures/screenshots/linux/mobile/blog/index/_pagination.png`
• `test/fixtures/screenshots/linux/mobile/blog/post.png`

Test plan

• macOS: `bundle exec ruby -Itest test/system/desktop_site_test.rb -n test_blog_index` → 1 run, 0 failures
• macOS: `bundle exec ruby -Itest test/system/mobile_site_test.rb -n "/test_blog_index|test_blog_post/"` → 3 runs, 0 failures
• Linux: same commands via Docker (`bin/docked t bundle exec ruby -Itest ...`) → 1 run desktop / 3 runs mobile / 0 failures
• Verified via fresh `git show HEAD:` lookup that committed baselines match captured screenshots

Process improvement (for separate session)

The visual regression gate is supposed to fire BEFORE merge per CLAUDE.md. The fact that 5 commits shipped through it without anyone updating baselines suggests the gate isn't actually blocking merges. Worth investigating in a follow-up — the rule's intent is to catch drift at the boundary, not 3 weeks later.

🤖 Generated with Claude Code

[coderabbitai]

Important

Review skipped

Review was skipped due to path filters

⛔ Files ignored due to path filters (7)

• test/fixtures/screenshots/linux/desktop/blog/index.png is excluded by !**/*.png
• test/fixtures/screenshots/linux/mobile/blog/index.png is excluded by !**/*.png
• test/fixtures/screenshots/linux/mobile/blog/index/_pagination.png is excluded by !**/*.png
• test/fixtures/screenshots/linux/mobile/blog/post.png is excluded by !**/*.png
• test/fixtures/screenshots/macos/desktop/blog/index.png is excluded by !**/*.png
• test/fixtures/screenshots/macos/mobile/blog/index.png is excluded by !**/*.png
• test/fixtures/screenshots/macos/mobile/blog/post.png is excluded by !**/*.png

CodeRabbit blocks several paths by default. You can override this behavior by explicitly including those paths in the path filters. For example, including **/dist/** will override the default block on the dist directory, by removing the pattern from both the lists.

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

Run ID: 0e7b3414-2298-4860-92a0-ba4d2ad3bd8e

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch chore/refresh-blog-baselines

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}