20.4Decision Framework

Choose Playwright CLI + Test Agents when:

• A coding agent with filesystem access drives your testing (the common case)
• Token efficiency matters (shared context with code editing)
• You want first-party plan → generate → heal workflow
• You need traces as an audit trail

Choose Playwright MCP when:

• The agent is sandboxed (no filesystem)
• Your IDE/agent integrates MCP natively
• Short exploratory sessions where context cost doesn't accumulate

Choose Stagehand when:

• You're building agent-powered product features in TypeScript
• You want intent-level control without selector or ref management
• Hosted browser infrastructure (Browserbase) fits your deployment

Choose browser-use when:

• Python is your primary language
• The UI defeats structural selectors (heavy canvas, custom components)
• You're doing research or building custom agent loops

Choose Selenium 4 when:

• Enterprise requirements (compliance, vendor support, exotic browsers)
• A large existing Selenium suite to maintain
• Multi-language team needs

Keep an eye on Vibium when:

• You care about W3C-standards-native tooling
• Its Cortex "app memory" ships — that's the feature nobody else has