[josephfusco]

End-to-End Tests is the slowest plugin workflow, so it decides when a PR gets its result. Over the last 300 runs its median was 3.3 minutes, but 8 of 36 E2E jobs took 10 to 44 minutes.

Each commit makes one change:

1. Playwright installs Chromium without --with-deps. Every slow run was apt downloading from azure.archive.ubuntu.com at 44 to 129 kB/s, while the runner image already had Chromium's libraries.
2. The Stop wp-env step is gone, since the runner is thrown away after the job anyway (13 seconds).
3. Plugin job checkouts skip the 14 wp-env/uploads.zip files, which are 590 MB of the 670 MB tree and only used by the example sites. Plugins with an examples/ folder already leave it out of their archive, so Plugin Check and the PR artifacts see the same files.
4. Detect jobs check out full history without file contents in place of git fetch --unshallow, which pulled every old version of those zips and took up to 47 seconds.

If a future runner image drops a library Chromium needs, the E2E job will fail at browser launch, and adding the package or restoring --with-deps fixes it. A plugin job that turns out to need an example zip would also fail until the sparse pattern changes.

These workflows only run on plugin changes, so this PR doesn't trigger them. #860 is a throwaway PR that runs them with these changes for timing.

Use of AI Tools

AI assistance: Yes
Tool(s): Claude Code
Model(s): Claude Opus 5.5
Used for: Assisting with analyzing GitHub Actions run history and making the workflow changes.

[josephfusco]

Timing from #860, which ran these changes on an hwp-previews comment edit, compared with the last 5 days of hwp-previews PR runs (median).

Workflow	Before	After
End-to-End Tests	294s	238s, 235s on rerun
Testing Integration	202s	152s
Plugin Check	176s	140s
Code Quality	86s	59s
Plugin Artifact for PR	77s	33s

E2E step	Before	After
Detect job	26s	6s
Checkout	11s	6s, 4s
Install Playwright browsers	46s (max 611s)	7s, 11s
Stop wp-env	13s	removed

The steps this PR touches went from about 96s to about 20s. E2E only dropped about 57s because Start wp-env (95s, 83s) and the tests (90s, 100s) ran slower than their usual 70s and 82s in both runs, and this PR doesn't change either step. Two runs can't show the tail, but the apt download behind every 10 to 44 minute run no longer happens.

[Copilot]

🟢 Approval recommended

No unresolved issues remain, and the reported timing run in PR #860 successfully exercised all five modified workflows.

No open issues 🎉

What changed in this PR

This PR reduces plugin CI runtime by avoiding large example uploads, redundant history downloads, and unnecessary E2E setup and cleanup.

Changes:

• Configure sparse, blobless checkouts across five plugin workflows, with full history for plugin detection.
• Remove redundant history fetching and disable rename detection in get_plugin_slug.sh.
• Skip Playwright system dependency installation and remove the final wp-env stop step.

File	Description
.github/​workflows/​plugin-check.yml	Optimize checkouts for plugin detection and validation.
.github/​workflows/​plugin-artifact-for-pr.yml	Optimize checkouts for plugin detection and artifact creation.
.github/​workflows/​e2e-test.yml	Optimize checkouts and reduce browser setup and environment cleanup.
.github/​workflows/​codeception.yml	Optimize checkouts for plugin detection and Codeception tests.
.github/​workflows/​code-quality.yml	Optimize checkouts for plugin detection and quality checks.
.github/​scripts/​get_plugin_slug.sh	Use checkout-provided history and avoid fetching blobs for rename detection.

_{Review effort: Balanced}