Run Playwright E2E tests safely (warns about dev server conflicts)
Scanned 9/3/2026
Install to Claude Code
npx -y skills add bloknayrb/tandem --skill e2e --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of E2e?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/bloknayrb-e2e)More formats (shields.io, HTML) on the badges page.
---
name: e2e
description: Run Playwright E2E tests safely (warns about dev server conflicts)
disable-model-invocation: true
---
# Run E2E Tests
Run the Playwright end-to-end test suite for Tandem.
## Ports (#1492)
The suite runs on **reserved ports** from `scripts/test-ports.ts` — Vite on 4573, backend on 4728 (ws) / 4729 (MCP). The product's 3478/3479 are untouched, so a running Tandem or `dev:server` coexists with an E2E run; you no longer need to quit anything.
**The reserved pair is *reserved*, not politely refused.** Playwright raises its terse "already used" error only for something answering 200–403 on :4729. Anything else holding the pair — a wedged stale E2E server, a non-HTTP process, or *anything at all* on :4728, which no Playwright check ever probes — is SIGKILLed by the E2E server's own boot (`freePort()`). For a stale E2E server that is desirable self-healing. Two guards still run in `scripts/e2e-guard.ts` (`globalSetup`): a fail-closed identity probe of :4729 (#1483), and a check that the Vite server actually serving the suite carries the harness env — a served client still baked to :3479 would drive the destructive suite into your REAL Tandem through the UI.
## Steps
1. Optionally check for a stale holder of the reserved pair:
```bash
if curl -sf http://127.0.0.1:4729/health 2>/dev/null; then
echo "Something is on :4729 (usually a stale E2E server) — the run will refuse"
echo "with Playwright's 'already used' error. Free it with:"
echo " fuser -k 4728/tcp 4729/tcp 4573/tcp"
fi
```
2. Run the tests:
```bash
npm run test:e2e
```
For interactive debugging mode:
```bash
npm run test:e2e:ui
```
3. If tests fail, check:
- `data-testid` attributes are present (kebab-case convention)
- Server started cleanly (check Playwright output for port conflicts)
- No stale `openDocuments` entries causing phantom tab removal
## Test Conventions
- Use `data-testid` selectors (kebab-case): `accept-btn`, `dismiss-btn`, `annotation-card-{id}`
- E2E tests live in `tests/e2e/`
- McpTestClient helper simulates MCP tool calls without Claude Code
Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.
No comments yet. Be the first to comment!