feat(webmcp): add page.webmcp API

[Pavel Feldman (pavelfeldman)]

No description provided.

[github-actions]

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

🟢 The one failure looks like a pre-existing flake

The only failure is reporter-html.spec.ts:3648 (speedboard) on macOS. It's an HTML-reporter UI test that this PR's page.webmcp changes don't reach, and it has flaked on the same bot before on another PR.

Details

This PR adds page.webmcp to the client, protocol, server, and Chromium code (crPage.ts, server/webmcp.ts) and refactors the MCP webmcp tool backend. It doesn't touch packages/html-reporter, the test runner, or the reporter tests. The latest report (run 36518047559) has 1 failure and 3 flaky tests, with 52,501 passing.

Pre-existing flake / infra

• [playwright-test] › reporter-html.spec.ts:3648 › merged › speedboard › clicking on label should not exit speedboard (macos-latest-node22): I checked this test's history across the test-results DB. On macos-latest-node22 it failed on the first attempt in one of 638 runs and passed on retry. That run was on feat(tracing): allow configuring the screenshots size #42283 (feat(tracing): allow configuring the screenshots size, sha 9e5bdac), which this PR can't have caused. The error there was a toMatchAriaSnapshot timeout because the speedboard view had been left. It has 0 failures in ~4,500 runs on the other bots (ubuntu/windows, all Node versions), so it only flakes on macOS and only rarely. The test works on the HTML report UI, which this PR doesn't touch.

Flaky, not triaged (passed on retry): browsercontext-page-event.spec.ts:160 (chromium arm) and two Firefox browsercontext-cookies-third-party.spec.ts Partitioned-cookie tests. None of them touch WebMCP.

Caveat: refreshing the test-results snapshot failed because the GitHub API returned an error, so this PR's own run isn't in the DB. The earlier flake above comes from the existing snapshot.

_{Triaged by the Playwright bot - agent run}

[Yury Semikhatsky (yury-s)]

move this to injected scripts

[github-actions]

Test results for "MCP"

3 failed
❌ [chromium] › mcp/sse.spec.ts:196 › sse transport browser lifecycle (persistent) @mcp-ubuntu-latest-chromium
❌ [firefox] › mcp/cli-core.spec.ts:97 › fill @mcp-windows-latest-firefox
❌ [firefox] › mcp/idle-timeout.spec.ts:19 › closes the browser after the idle timeout and relaunches it on the next call @mcp-windows-latest-firefox

8826 passed, 1480 skipped

---

Merge workflow run.

[github-actions]

Test results for "tests 1"

3 flaky

⚠️ [chromium-library] › library/video.spec.ts:762 › screencast › should work with video+trace `@chromium-ubuntu-22.04-node20`
⚠️ [firefox-library] › library/browsercontext-cookies-third-party.spec.ts:257 › third party 'Partitioned;' cookies `@firefox-ubuntu-22.04-node20`
⚠️ [firefox-library] › library/browsercontext-cookies-third-party.spec.ts:470 › top level 'Partitioned;' cookie and same origin iframe `@firefox-ubuntu-22.04-node20`

52520 passed, 1270 skipped

---

Merge workflow run.

[github-actions]

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

🟢 All three MCP failures look like pre-existing flakes

Each failing test has also failed, on the same bot, on main pushes or unrelated PRs.

Details

The only MCP-side change in this PR is in tools/backend/tab.ts. It calls frame.webmcp.enable() per frame and catches any error. That path runs in every MCP test, and 8826 of them passed, so a regression there would show up much more widely than three one-off failures on separate bots. I couldn't get the error text out of the job logs, so this verdict rests on history. The results DB snapshot also couldn't be refreshed (GitHub API 500), so the latest runs aren't in it.

Pre-existing flake / infra

• [chromium] › mcp/sse.spec.ts:196 › sse transport browser lifecycle (persistent) (mcp-ubuntu-latest-chromium) failed 3 of 199 runs on this bot, including main pushes (36070734686, 36118130299) and feat(chromium): roll to r1247 #42915. The same test also fails or flakes on webkit, firefox and msedge across Ubuntu, macOS and Windows.
• [firefox] › mcp/cli-core.spec.ts:97 › fill (mcp-windows-latest-firefox) failed 4 of 370 runs on this bot, including main pushes (34656353189, 33644076891), feat(evaluate): add "world" option to evaluate methods #42635 and test(webkit): reenable WebSocket handshake HAR coverage #42513.
• [firefox] › mcp/idle-timeout.spec.ts:19 › closes the browser after the idle timeout and relaunches it on the next call (mcp-windows-latest-firefox) failed 4 of 110 runs on this bot, in Fix spelling of restoring in credentials docs #42718, chore(coverage): istanbul instrumentation for the trace viewer and html reporter #42785, feat(screencast): style action decorations, keep cursor visible #42758 and devops(bidi): publish BiDi reports to a git branch instead of Azure #42760. It also failed on main pushes on mcp-macos-latest-chrome and was rescued by a retry on other firefox and chrome bots.

_{Triaged by the Playwright bot - agent run}