← Back to the library
CodingComposed workflow

Build it. See it. Fix it.

Pair a coding harness with Playwright CLI and skills, or MCP when you need an ongoing browser session. Inspect the local app with an isolated browser profile.

What this stack does

A frontend with a visual feedback loop

Components and their roles

OpenCode / coding harnessCode and task loop ↗Playwright CLI + SkillsCommand-based browser feedback ↗Playwright MCP (alternative)Persistent browser tool session ↗

How to get started

  1. Choose CLI and skills for command-based checks, or MCP for an ongoing browser tool session.
  2. Use a separate automation browser and fresh profile; build one screen and inspect its behavior.
  3. Check phone and desktop layouts, keyboard use and key interactions; save regression tests for important flows.

Limitations

Browser access can reach accounts and local resources. Use task-specific permissions; this recipe does not establish a quality advantage over other stacks.

Costs and licensing

Tooling licenses and provider costs differ; consult official terms.

Evidence level

Documentation reviewed · not hands-on tested. This is an editorial combination, not a packaged product or a benchmark result.

Sources

Primary sourceCurrent OpenCode documentation ↗Primary sourceMicrosoft Playwright CLI and skills ↗Primary sourceMicrosoft: CLI and MCP tradeoffs ↗

Content checked: 2026-11-04. This page reflects the current validated published catalog. Update method and live status ↗