Menú
De forma predeterminada está seleccionado el prompt que primero revisa el origen. Puedes cambiar a un comando directo o descargar una copia local.
Lee SKILL.md y los archivos complementarios que muestra SkillsMP antes de decidir si quieres instalarlo.
Instalar con Codex o Claude Copia este prompt, pégalo en Codex, Claude u otro asistente, y deja que revise la página de la skill y la instale por ti.
Un comando directo omite el prompt de revisión. Revisa el origen antes de ejecutarlo.
npx skills add https://github.com/wp-media/maestro --skill e2eEl comando permanece en una sola línea. Desplázate horizontalmente para revisarlo antes de copiarlo.
¿Prefieres una copia local? Descarga los archivos que SkillsMP tiene disponibles ahora.
Basado en la clasificación ocupacional SOC
Mostrando SKILL.md
| name | e2e |
| description | Run E2E smoke tests (basic) or full acceptance + regression suite (extended). |
Read .claude/maestro.json to resolve project-specific values:
{TEMP_ROOT} = .ai.temp_root{REPO} = .ai.repo{SLUG} = .ai.slug{E2E_URL} = .ai.e2e.local_url{E2E_BOOT} = .ai.e2e.boot_cmd{E2E_SETTINGS} = .ai.e2e.settings_path{E2E_CI} = .ai.e2e.ci_integrationThis skill provides end-to-end test execution at two tiers. The tooling is the same (Playwright MCP + curl + WP-CLI); the difference is scope and depth.
The local WordPress environment is at {E2E_URL} (admin / password). Boot
or restart it with {E2E_BOOT}.
Purpose: behavioral verification and smoke tests. Fast enough to fit inside a planning agent's execution window.
Invokers:
grooming-agent — verify behavioral assumptions about the current system before
writing the spec. Use to confirm: does the current feature behave as described in the
issue? What does the current API or AJAX endpoint return for the scenario being changed?| You'll be tempted to say | Why you can't |
|---|---|
| "The environment probably isn't up, I'll skip" |
Run {E2E_BOOT}. It's idempotent. If it fails, log SKIP with the reason — do not silently omit the step. |
| "The change is backend-only, no need to smoke it" | The primary happy path must be verified. A backend change with no observable behavior change still needs a confirming assertion. |
| "I already read the code, I know it works" | "Seems right" never closes a task. Run the scenario. |
| "One scenario is too slow for this stage" | Basic tier is exactly one primary scenario. The cost is acceptable. |
Boot the environment (idempotent — safe to run if already up):
{E2E_BOOT}
If the script exits non-zero, set status: "SKIP", note the reason, and do not block
the pipeline.
Run the primary happy path scenario from the spec or grooming plan.
Backend / AJAX / REST:
# Public REST or AJAX
curl -s -X POST {E2E_URL}/wp-admin/admin-ajax.php \
-H "Cookie: $(cat .wp-session-cookie 2>/dev/null)" \
-d 'action=<action>&nonce=...'
# Cache headers
curl -sI {E2E_URL}/ | grep -E '(x-cache|cf-cache)'
# WP-CLI inside the dev environment
bin/wp <command>
Browser (settings page, dashboard notices, interactive UI):
Use the Playwright MCP directly for basic-tier smoke. Do not delegate to
e2e-qa-tester at this tier (that is the extended tier path):
mcp__playwright__navigate({ url: "{E2E_URL}/wp-login.php" })
# login
mcp__playwright__fill({ selector: "#user_login", value: "admin" })
mcp__playwright__fill({ selector: "#user_pass", value: "password" })
mcp__playwright__click({ selector: "#wp-submit" })
# primary scenario
mcp__playwright__navigate({ url: "{E2E_SETTINGS}" })
mcp__playwright__snapshot({})
# Inspect snapshot output to confirm expected element/text is present
Take at most 1–2 screenshots if helpful, but do not publish them at this tier.
Report:
{
"status": "PASS|FAIL|SKIP",
"scenarios_tested": ["Settings page loads without errors after enabling X option"],
"details": "Logged in as admin, navigated to {E2E_SETTINGS}, confirmed no JS console errors and X toggle present"
}
SKIP: {E2E_BOOT} failed or environment unreachable. Record reason. Do not block
the pipeline.
e2e-qa-tester)Purpose: full acceptance criteria coverage, regression testing, edge cases, visual comparison, and Playwright spec authoring with screenshot evidence.
Invoker: qa-engineer only.
Execution: the qa-engineer agent delegates browser flows to the e2e-qa-tester
sub-agent, which handles Playwright MCP driving, temporary spec authoring under
.e2e-temp/, screenshot publishing via the commit-SHA method, and clean-up.
The qa-engineer agent itself handles:
For details, read:
.claude/agents/qa-engineer.md — strategy selection and report format.claude/agents/e2e-qa-tester.md — browser flow execution, spec authoring, screenshot publishingThe extended tier writes Playwright specs to .e2e-temp/ (gitignored, never committed when {E2E_CI} is false)
and screenshots to .e2e-screenshots/. Screenshots are published to a public GitHub Gist (gh gist create --public) to get permanent, publicly accessible raw URLs — no commits to the PR branch. Gist raw URLs never 404 in PR comments, unlike commit-SHA-based URLs.
| Invoker | Tier | Purpose |
|---|---|---|
grooming-agent | Basic | Verify a behavioral assumption before writing the spec |
qa-engineer | Extended | Full acceptance criteria + regression + screenshots |
{E2E_BOOT} is idempotent. Always run it before testing — don't pre-check whether the environment is up.admin / password.{E2E_SETTINGS}.curl -s -o /dev/null -w "%{http_code}" {E2E_SETTINGS}
{E2E_CI} is true, the project maintains a permanent E2E suite — e2e-qa-tester will commit spec files rather than delete them.