skill
Pick browser automation for web/Electron: CI E2E, scripted flows, scraping, visual regression, exploratory QA, persona walks, monitoring, or browser agents. Deterministic Playwright first; harden exploratory findings into repeatable tests. Use for "automate the browser", "test this web app", "test this electron app", "Playwright or Stagehand", "scrape this site", "browser agent", "visual regression", "E2E tests". Trigger: /browser.
/browser
Source: skills/browser/SKILL.md
Pick browser automation for web/Electron: CI E2E, scripted flows, scraping, visual regression, exploratory QA, persona walks, monitoring, or browser agents. Deterministic Playwright first; harden exploratory findings into repeatable tests. Use for "automate the browser", "test this web app", "test this electron app", "Playwright or Stagehand", "scrape this site", "browser agent", "visual regression", "E2E tests". Trigger: /browser.
Support primitive in the Harness Kit operating loop.
This generated excerpt gives readers the beginning of the live primitive contract before they jump to GitHub.
Browser automation is a **testing pyramid**, not a single tool. Pick the
layer first, then the tool for that layer.
## Delegation Floor
Delegation floor applies for substantive browser strategy, exploratory QA,
persona walks, visual regression design, or hardening findings into tests:
probe the roster first; dispatch two or more providers; direct solo only for
mechanical browser commands, emergency preservation, user-forbidden
delegation, or fewer-than-two-providers cases. See
`harnesses/shared/AGENTS.md` (Roster).
Local lane guidance: Use specialized lanes for tool/stack selector,
persona explorer, deterministic-test author, visual/evidence reviewer, and
flake-risk critic. Runtime browser agents are evidence inputs; the lead owns
which findings become committed deterministic tests.
## The Pyramid
| Layer | Purpose | Tooling |
|-------|---------|---------|
| **4. Continuous QA** | scheduled agents, autonomous bug filing, synthetic monitoring against staging/prod | Custom Browser Use / Stagehand loops, bugAgent, QA.tech, Mabl, supaguard |
| **3. Exploratory / Persona-driven** | cold-start exploration with a persona, charters, SBTM/PROOF reports, UX-gap discovery | Browser Use, Stagehand `agent()`, agent-browser, custom persona harnesses |
| **2. Hybrid / AI-assisted** | Playwright body + AI for fragile steps, self-healing, AI-authored tests committed to repo | Playwright 1.56+ Planner/Generator/Healer, Stagehand atomic primitives, QA Wolf / Octomind (generate code), visual regression |
| **1. Deterministic Playwright E2E** | critical user journeys, CI gate, 99%+ reliable, regression floor | Raw Playwright code, Playwright Test, fixtures, `getByRole` |
**Findings flow down, not up.** The exploratory layer is a *discovery
instrument*. When an agent persona finds a real bug or UX gap, harden
...