Atlas › 01 Agent Skills for Browser Automation › The Playwright CLI: Browser Control for Coding Agents☰ Read as one page
The Playwright CLI: Browser Control for Coding Agents
4.1Why This Chapter ExistsIn early 2026 Microsoft shipped a standalone command-line tool — @playwright/cli — built specifically for AI coding agents, and then did…4.2Installation and Skill SetupThe --skills flag generates a SKILL.md file — the same mechanism you learned in Foundations. The agent reads one markdown file to learn…4.3The Disk-First ArchitectureThe key design decision — and the reason for the token savings — is where browser state goes:4.4Element References, Not SelectorsSnapshots are compact YAML accessibility summaries with stable references:4.5Command SurfaceUnder the hood this is the same battle-tested Playwright engine: auto-waiting, actionability checks, cross-browser support — you give up…4.6Named SessionsParallel browser contexts come free via named sessions:4.7When CLI, When MCPMicrosoft's own guidance reduces to one question: does the agent have filesystem access?4.8Interview Talking Point"We drive browsers through the Playwright CLI rather than Playwright MCP. Microsoft's own benchmark shows about 4x fewer tokens — roughly…