e2e-runlisted

# e2e-run · Execute Live E2E Testing and Collect Evidence This skill defines how to execute live end-to-end testing and collect physical evidence: test cases originate from `docs/e2e/<client>.md`, evidence is recorded in `docs/e2e/TestResult/<client>.md` (refer to its README for evidence guidelines), and test suite scripts are located in `scripts/`. ## Mandatory Prerequisites 1. **Verify Guard Build**: Ensure the guard under test is built from the latest codebase (`npm run build` has been executed, `lib/` is fresh). 2. **Matrix and Realflow Regression**: Execute `npm run test:git-matrix` (the 135-case exhaustive Git command decision matrix) and `npm run test:realflow` (full feature branch lifecycle flow), asserting that all tests pass. 3. **Controlled Sandboxes**: Run client tests strictly within temporary directories (e.g., `/tmp/e2e-<client>-repo` with `master=integration`, `beta=preview`, `(fix|task)/*=feature`, local bare remotes, and local config). **Never execute commands with successful push side-effects against real remotes.** 4. **Client Credentials**: Smoke-test each client with a simple prompt (e.g., `codex exec "Reply with exactly: OK"`, `claude -p "Reply with exactly: OK"`, `pi --mode json ... "PI-OK"`, `opencode run "OK"`). When running in restricted sandbox environments, mirror credentials using platform-specific XDG/temporary directory variables. ## Execution Workflow 1. **Wire Configuration** (stdin-hook clients): Run `gitflow-guard wire --client <x> --p