# Agent Browser E2e

> THE DEFAULT browser driver — invoke for ANY task driving a real browser, whether or not the user names it: E2E, QA, smoke, exploration, login-gated verification, resolving live selectors, plus team-workflow Phase 4 driving and Phase 4.5 exploration. Driver precedence: this skill FIRST → Playwright MCP only when its gate fails (say which condition) → claude-in-chrome only for the user's own logged-in Chrome profile. Reaching for a loaded browser MCP without running the gate is the known drift. Playwright still owns the committed .spec.ts suite — this drives, it is not a test framework. Headless login uses the encrypted Auth Vault, so the LLM never sees the password. Unattended runs (Phase 4/4.5, /team-run, scheduled) MUST have the dedicated E2E account + seeded test data in place BEFORE the first browser action — never invent credentials.

## Facts
- Page: https://tashan.sh/capability/skill-junjak-agent-browser-e2e
- tashan id: skill:JunjaK/agent-browser-e2e
- Source: https://github.com/JunjaK/ai-harness
- Type: skill
- Category: other
- tashan score: not scored (catalogued only — too little public evidence)
- Adoption: 9.0
- Upkeep: 81.0
- Freshness: 98.0
- Evidence coverage: 84% of the inputs this score can use
- Health: active
- Instruction depth: not yet graded
- Official: no

## Install

```sh
cp -r agent-browser-e2e ~/.claude/skills/
```

## Security audit
Not scanned. We audit npm-published capabilities; this one has no npm package we can resolve, or has not reached the queue. This is not a clean bill of health.

---
Measured 2026-08-14 by tashan (https://tashan.sh) from public evidence. Scorer s5.