chore: remove experimental component testing

[Pavel Feldman (pavelfeldman)]

Summary

• Delete @playwright/experimental-ct-{core,react,react17,vue} packages, tests/components/, the CT specs and the components CI workflow.
• Remove the '@playwright/test': { babelPlugins } hook that injected tsxTransform, plus the setTransformData/getUserData cache side channel and resolveCtDirs() it fed.
• Rewrite the four non-CT tests that happened to use CT fixtures (test-server find-related-files/clear-cache, two aria-snapshot rebaseline tests) so their coverage survives.
• Point testing-library.md at the story gallery, and note in test-components.md / the migration skill that the packages are gone (pin 1.62 while porting).

jsx-runtime stays for now — it is inert without CT, but dropping it would break JSX in .tsx test files, so it deserves its own breaking change.

[github-actions]

Test results for "tests 1"

1 failed
❌ [playwright-test] › reporter-html.spec.ts:3402 › created › speedboard › clicking on label should not exit speedboard @macos-latest-node22

4 flaky

⚠️ [chromium-library] › library/video.spec.ts:495 › screencast › should capture static page in persistent context Radoslav Kirilov (@smoke) `@frozen-time-library-chromium-linux`
⚠️ [chromium-library] › library/video.spec.ts:736 › screencast › should work with video+trace `@realtime-time-library-chromium-linux`
⚠️ [firefox-page] › page/page-emulate-media.spec.ts:144 › should keep reduced motion and color emulation after reload `@firefox-ubuntu-22.04-node20`
⚠️ [playwright-test] › ui-mode-trace.spec.ts:827 › should update state on subsequent run `@windows-latest-node22`

51005 passed, 1190 skipped

---

Merge workflow run.

[github-actions]

Hi, I'm the Playwright bot and I took a first look at the failing CI.

🟢 The one failure is a pre-existing flake — this PR is clear

reporter-html.spec.ts:3402 › speedboard › clicking on label should not exit speedboard fails on macos-latest-node22, but this PR only deletes experimental component testing and doesn't touch the HTML reporter it exercises.

Details

Overall: the diff is confined to the @playwright/experimental-ct-* packages, tests/components/, the components workflow, a few docs, and package manifests. It doesn't touch the HTML reporter, packages/web, or the playwright-test reporters — so nothing in it reaches the speedboard flow.

Pre-existing flake / infra

• [playwright-test] › reporter-html.spec.ts:3402 › created › speedboard › clicking on label should not exit speedboard — pre-existing flake. Across the test-results DB this test is green in 4386 of 4387 runs; its single other failure was on PR chore(deps-dev): bump fast-uri from 3.1.3 to 3.1.5 #42112 (same macos-latest-node22 bot), a SHA this PR can't be responsible for, with an aria-snapshot 75s timeout signature — and two more runs were rescued by a retry. The PR doesn't touch the HTML reporter, so it can't be the cause here.

Not real failures

• The 4 remaining results in the report are flaky (passed on retry): library/video.spec.ts:495, library/video.spec.ts:736, page/page-emulate-media.spec.ts:144, and ui-mode-trace.spec.ts:827. None are in code this PR changed.

Worth noting: earlier merged reports for this PR ended cancelled, so their counts were partial — the verdict above is based on the latest complete report.

I'm a first pass, so I can't rule out a rerun surfacing something new, but the evidence points squarely at flake.

_{Triaged by the Playwright bot - agent run}

[github-actions]

Test results for "MCP"

5 failed
❌ [firefox] › mcp/cli-core.spec.ts:394 › --raw on command without output @mcp-windows-latest-firefox
❌ [firefox] › mcp/cli-devtools.spec.ts:84 › request shows full request and response details @mcp-windows-latest-firefox
❌ [firefox] › mcp/cli-devtools.spec.ts:115 › per-part commands extract individual parts @mcp-windows-latest-firefox
❌ [firefox] › mcp/cli-devtools.spec.ts:138 › request* and response* commands support --filename @mcp-windows-latest-firefox
❌ [firefox] › mcp/cli-session.spec.ts:130 › workspace isolation - sessions in different workspaces are isolated @mcp-windows-latest-firefox

8058 passed, 1284 skipped

---

Merge workflow run.

[github-actions]

Hi, I'm the Playwright bot and I took a look at the failing CI.

🟢 All five MCP failures are pre-existing flakes — this PR is clear

The latest MCP report lists 5 failed tests, all on mcp-windows-latest-firefox. This PR only removes experimental component testing and touches no MCP or CLI code, and every one of these tests has failed the same way on other PRs/main.

Details

Overall: the diff is confined to @playwright/experimental-ct-*, tests/components/, the components workflow, a few docs, and package manifests. It doesn't touch mcp/, the CLI, or any server code these tests exercise — so nothing in it reaches these Firefox MCP CLI tests.

Pre-existing flake / infra — every failure recurs on the same mcp-windows-latest-firefox bot on SHAs this PR can't be responsible for (704 runs of each in the results DB):

• [firefox] › mcp/cli-core.spec.ts:394 › --raw on command without output — also failed on PR [DO NOT LAND] roll ffmpeg to 1012 #42067 (chore-roll-ffmpeg), same bot.
• [firefox] › mcp/cli-devtools.spec.ts:84 › request shows full request and response details — also failed on PR Revert "feat(routeFromHar): add interceptAPIRequests option (#41294)" #41964 and PR test(filechooser): allow exact one second lastModified rounding #42145, same bot.
• [firefox] › mcp/cli-devtools.spec.ts:115 › per-part commands extract individual parts — also failed on PR chore: mark v1.62.1 #42020 and on main (run 30469109080), same bot.
• [firefox] › mcp/cli-devtools.spec.ts:138 › request* and response* commands support --filename — also failed on PR test: roll stable-test-runner to 1.63.0-alpha-2026-08-03 #42091 (roll-stable-test-runner), same bot.
• [firefox] › mcp/cli-session.spec.ts:130 › workspace isolation - sessions in different workspaces are isolated — also timed out on PR fix(electron): close pages from the main process to avoid Target.closeTarget hang #41695 (test-electron-teardown-hang), on mcp-macos-latest-chromium.

The clustering on one Firefox/Windows MCP bot, all in code the PR leaves untouched, is the shape of pre-existing infra flake, not a regression.

Worth noting: earlier merged reports for this PR ended cancelled, so their counts were partial. A separate "tests 1" report's lone reporter-html.spec.ts:3402 failure was also a pre-existing flake in code this PR doesn't touch.

I'm a first pass, so I can't rule out a rerun surfacing something new, but the evidence points squarely at flake.

_{Triaged by the Playwright bot - agent run}