agentqx

<p align="center"> <h1 align="center">agentqx</h1> <p align="center"> <strong>Give it a URL. Go to sleep. Wake up with a battle-tested E2E test suite.</strong> </p> <p align="center"> A CLI and MCP server that builds full Playwright E2E test suites for any web app. You provide a URL; it explores the site, generates tests, runs them, fixes failures, and adds tests for uncovered pages—until every test passes or it hits safety limits. </p> <p align="center"> <a href="https://www.npmjs.com/package/agentqx"></a> <a href="https://www.typescriptlang.org/"></a> <a href="https://github.com/sijeeshmiziha/agentqx/blob/main/LICENSE"></a> <a href="https://github.com/sijeeshmiziha/agentqx/pulls"></a> </p> </p>

---

1. Explore — Crawls the URL and builds a site map (pages, structure, screenshots).
2. Generate — Uses an LLM to create Playwright tests from the site map.
3. Run — Executes the test suite.
4. Fix — If any test fails, sends failures + code to the LLM and applies fixes.
5. Repeat — Re-runs tests, fixes again, and adds tests for pages that aren’t covered yet. Stops when all tests pass or when max iterations / stagnation limits are reached.

No extra complexity: no coverage scores to hit, no “keep vs discard” decisions—only pass or fail, and the loop keeps going until everything passes.

---

• Node.js ≥ 20
• Playwright — npx playwright install chromium
• One LLM provider — set the matching env var or use `--prov