workflow-engineer-diagnose-e2elisted

# workflow-engineer-diagnose-e2e The **diagnose half** of the Pass-E2E phase. Run the slice's E2E specs against a real stack and report *what is failing and why* — grouped by shared root cause — so the calling `implement-slice` workflow can dispatch one focused engineer fix per group. This step **writes no production code and no specs**. It reproduces failures and categorizes them; the actual TDD fix is a separate dispatch (`workflow-engineer-fix-e2e`), one per correlated group, run serially on the shared slice worktree. Keeping diagnose fix-free is what lets each fix run in a fresh, narrow context scoped to one root cause instead of one mega-context juggling the whole suite. The single exception to "no writes": the mandatory `origin/main` integration merge (step 2) is committed and pushed, exactly as the old combined skill did — integration must persist so the fixers and the next diagnose round see it. ## When to activate Activate this skill whenever: - The dispatch prompt opens with `Diagnose E2E acceptance for slice #<n>`. - The user types `/workflow-engineer-diagnose-e2e`, or "diagnose the slice's E2E failures". Do NOT activate to FIX a failure (that is `workflow-engineer-fix-e2e`), to author / modify E2E specs (`workflow-e2e-author` / `workflow-e2e-fix`), to fix reviewer findings on a slice (`workflow-engineer-fix-slice`), or to fix a PR (`workflow-engineer-fix-pr`). ## Input contract Read the slice issue #<n> body. The slice's Acceptance criteria (EARS) plus ea