By name: fm-e2e
description: "Use after fm-verify to run the Playwright E2E gatekeeper on a migrated page — realizes the planned scenarios, dual-runs against the legacy app for behavior parity, and runs transactional flows against staging gateways."
argument-hint: " [--app pc|mobile|hana]"
user-invocable: true
allowed-tools: Read, Write, Edit, Glob, Grep, Bash, Agent
E2E Gate (Playwright)
The functional gatekeeper between fm-verify and fm-parity. No route flip until this passes.
All user-facing output in workingLanguage.
Instructions
Step 0: Config & prerequisites
Read config (absent → run fm-init; stop). Resolve app, appDir, targetDir, legacyDir,
workingLanguage, and stagingConfig (payment-gateway test endpoints). Require the page at
verified in tracker.json and migration-plan.json with e2eScenarios (else point to
fm-verify/fm-plan).
Step 1: Ensure Playwright run permission
The runner executes as a sub-agent, so session approvals do not transfer. Ensure
.claude/settings.json permissions.allow includes the Playwright command
(e.g. Bash(npx playwright *)). If missing, add it (Read-Modify-Write the settings file) and
note it in the report.
Step 2: Lock
Acquire docs/migration/{app}/{page}/.lock (stale after 30 min).
Step 3: Run the gate
Launch e2e-test-runner (Agent) with only its params: app, page, planPath, targetDir,
appDir, legacyDir/legacy base URL, stagingConfig, outPath =
docs/migration/{app}/{page}/e2e-report.json, workingLanguage. The skill starts/stops any dev
server the runner needs.
Step 4: Record
Read e2e-report.json. Update tracker.json (Read-Modify-Write):
result: pass→apps[app].pages[page].status = "e2e-passed".result: fail→e2e-failed. Release the lock.
Step 4b: Codex audit (advisory) — see CLAUDE.md → "Codex Independent Audit"
If codexAudit is enabled and Codex is available, after the lock is released spawn codex-auditor
(Agent) for the e2e stage (params: app, page, stage="e2e", appDir, legacyDir,
e2eReportPath + planPath, outPath = docs/migration/{app}/{page}/codex-audit.json,
workingLanguage). The Codex cross-check here targets false passes — whether the scenarios
truly cover legacy parity. Advisory — never changes the page status. Surface its verdict below.
Step 5: Report
In workingLanguage: scenarios run (msw vs staging), pass/fail with evidence, legacy dual-run
parity, the Codex audit verdict (advisory), and the next step — on pass
/frontend-migration-plugin:fm-parity {page}; on fail /frontend-migration-plugin:fm-fix {page}
(auto-detects e2e-fix mode).