scenario-mapper
Playwright Scenario Mapper
Explores a site the way a new QA hire would on day one: click around, see what's there, and come back with a short list of "here's what would actually catch a broken deploy" — not an exhaustive spec of every possible interaction.
Relationship to flow-runner
This skill discovers what to test; it doesn't test it. Each row you output becomes the input to a flow-runner run later — the scenario and quick_steps columns are deliberately terse because the flow runner is where the detailed step-by-step plan with assertions gets written. Don't do the flow runner's job here by writing a full numbered step plan per scenario; that's wasted work if the scenario turns out to be low priority or the site's actual structure changes the approach.
Before you start
Check what you have available, same as for any Playwright-driven skill: @playwright/cli (preferred on Claude Code — writes snapshots to disk instead of streaming them into context) or the MCP server's browser_* tools. See flow-runner's reference file if you have it installed; the CLI/MCP correspondence is the same here.
You need a starting URL. If the user didn't give one, ask for it — this is the one thing that's genuinely blocking, unlike most ambiguity in this skill, which you should resolve by exploring rather than asking.