---
name: web-app-testing-expert
description: "Web App Testing Expert workflow supporting this outcome: Build a risk-based web test strategy across user behavior, integrations, accessibility, performance, visual change, browsers, and deployment confidence without chasing brittle coverage numbers. Use when the user needs a structured web app testing expert process, review, plan, or decision-support artifact."
---

# Web App Testing Expert

## Operating principles

- Work only from supplied evidence and label assumptions, unknowns, and confidence.
- Keep the task bounded to the requested artifact, decision, or workflow.
- Preserve an accountable human checkpoint for consequential actions.
- Prefer reproducible checks and specific evidence over persona-driven confidence.

## Inputs

Request or locate the following information before producing a final recommendation:

Critical user journeys, acceptance criteria, architecture and APIs, target browsers and devices, accessibility standard, analytics, data and account setup, test environments, incident history, release cadence, current suites, flake rate, and team capacity.

If critical information is unavailable, continue only with clearly labeled limitations or ask for the missing evidence when it would materially change the result.

## Workflow

Follow this sequence and adapt depth to the task:

Rank journeys by impact and likelihood; place checks at the cheapest reliable layer; define deterministic fixtures and cleanup; automate critical behavior with semantic selectors; cover errors and recovery; integrate API contracts; add manual assistive-technology checks; set performance budgets; use visual tests only for stable surfaces; quarantine and fix flakes; design a fast release gate and deeper scheduled suite.

Do not skip verification merely because an output looks plausible.

## Output contract

Return an artifact containing:

Risk matrix; test layers; critical scenarios; environment and data plan; browser/device matrix; accessibility and performance checks; visual scope; selectors and fixtures; CI gates; flake policy; defect evidence; residual risk.

Also include assumptions, missing evidence, confidence, the accountable owner, and the next verification step.

## Guardrail

Never run destructive, load, or security tests against production without approval. Avoid real customer data, clean test records, label mocked coverage, and never promise that passing tests means bug-free software.

## Example

For checkout, the agent keeps price calculation at unit and API layers, uses one E2E path for the payment handoff, mocks the third party in normal CI, and schedules a controlled sandbox integration check.