docs(cli): cover Python pytest CLI debug in skills and coding-agents guide

[Simon Knott (Skn0tt)]

Extends the shared playwright-cli skill and the coding-agents guide for Python's pytest --playwright-debug=cli -s path (same attach loop as JS --debug=cli), and documents python -m playwright cli as a bundled entry point.

Do not merge until pytest-playwright with --playwright-debug=cli is released - that's microsoft/playwright-pytest#315.

Fixes #38610

[github-actions]

Test results for "MCP"

2 failed
❌ [chrome] › mcp/annotate.spec.ts:57 › should capture multiple screenshots in one annotation @mcp-windows-latest-chrome
❌ [chrome] › mcp/annotate.spec.ts:110 › should abort annotation when last screenshot is removed @mcp-windows-latest-chrome

8061 passed, 1284 skipped

---

Merge workflow run.

[github-actions]

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

🟢 Both failures are pre-existing flakes — this PR is clear

This is a docs-only change (skill markdown + the getting-started CLI guide), and the two failures are in mcp/annotate.spec.ts, which the diff doesn't touch. Both show a long history of flaking on SHAs unrelated to this PR.

Details

The diff only edits docs/src/getting-started-cli.md and two files under packages/playwright-core/src/tools/skills/playwright-cli/. No product code path exercised by the MCP annotate flow is affected, so there's no route from this change to either failure.

Pre-existing flake / infra

• [chrome] › mcp/annotate.spec.ts:57 › should capture multiple screenshots in one annotation @mcp-windows-latest-chrome — bimodal across the test-results DB. On chrome it failed 10 of 693 runs (~1.4%) and passed the rest, and it also flips on firefox (4/678), webkit (2/690), chromium (1/685) and msedge (1/442) — all on SHAs unrelated to this PR.
• [chrome] › mcp/annotate.spec.ts:110 › should abort annotation when last screenshot is removed @mcp-windows-latest-chrome — same picture: failed 8 of 693 chrome runs (~1.2%), passed 685, and flips on firefox (13/678), chromium (5/685) and webkit (3/690) elsewhere.

I'm a first pass, not the final word, but the evidence points the same way on both: unrelated flakes, safe to re-run.

_{Triaged by the Playwright bot - agent run}