Vedha

Your Playwright tests pass over silent 500s. Change one import.

npm install -D @vedhajs/playwright-test

import { test, expect } from "@vedhajs/playwright-test"; // was "@playwright/test"

_{Run this demo: npx tsx examples/guard-demo/run.ts}

The page says "Order placed", so the test passes. But the server answered POST /api/orders with a 500 and stored nothing. With the import changed, your tests run exactly as before, and a test whose own checks pass still fails when the app answered a request with a 5xx or a write failed outright. No rewrite, no new assertions. Details and options: guard mode. For CI, see examples/ci/.

And when an AI agent does the testing

An agent that trusts the screen makes the same mistake: it clicks "Save", sees "Saved" and reports success. Vedha's MCP server, CLI and SDK run on Playwright and check that each action really happened (page, network, and a read-back of stored data) before they report a pass.

npx -y @vedhajs/mcp --allow-domain localhost     # MCP server for your agent
npm install -D @vedhajs/cli                      # or the CLI / SDK for checked steps

_{Agent-side demo (Playwright PASS vs Vedha FAILED on the same silent 500): npx tsx examples/demo/run.ts}

Head to head: AI agent testing a web app

Same model, same tasks, written before any run on them. Playwright MCP reports success whenever the page looks right; Vedha also checks what the server did.

The lead held on every held-out set: 10 vs 4 of 14 (set 2), 15 vs 10 of 18 (set 3) and 16 vs 10 of 18 (set 4), or 41 vs 24 caught overall.

_{Fine print: LLM results are from one free model (minimax/minimax-m3.1-flash) at 3 trials per cell. When an expert writes every check by hand, Vedha and Playwright tie; Vedha's lead is with typical checks and with AI agents. All numbers}

... View full README on GitHub