skills.rest
apireno
apirenoCOMMUNITY
Explore all skills by this author →
GitHub

qa-ux

Drives live applications across browser, CLI, and MCP surfaces to produce defect reports and UX flow-graphs.

1|Updated Jun 13, 2026
Search Tags:#browser automation#mcp#qa testing#sprint acceptance#ux verification#defect report#flow-graph
One-click install
npx skills add https://github.com/apireno/agent-workflow-template --skill qa-ux-apireno
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: qa-ux
Source: https://github.com/apireno/agent-workflow-template/tree/main/.claude/skills/qa-ux
Command: npx skills add https://github.com/apireno/agent-workflow-template --skill qa-ux-apireno

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? User-facing sprints often ship without systematic end-user verification, letting UX defects and broken flows slip through code review. This Skill acts as a skeptical end-user that exercises the running product before sprint acceptance, producing evidence-backed defect reports instead of ad-hoc manual clicking. ## Core Features & Use Cases - Two-phase QA workflow: --mode plan authors a qa-plan.md at sprint-planning time with HARD (deterministic) vs SOFT (semantic) assertions, hero shots, and in-scope surfaces; --mode drive is the accept-time gate that exercises the live app against that plan. - Multi-surface driving: tests browser surfaces via DOMShell MCP (isolated tab lanes), CLI tools via Bash with exit-code assertions, and MCP servers for tool-description fidelity and error recovery. - Structured outputs: generates qa-report.md with severity-tagged findings (BLOCKER/MAJOR/MINOR/NOTE) routed by fault-domain, a UX flow-graph (JSON + Mermaid) diffed against prior sprints, and redacted provenance-stamped hero assets. - Use Case: Before accepting a sprint that delivered a new checkout flow, run the drive mode to navigate the live app as a user, capture screenshots of the checkout hero feature, log a MAJOR defect where the confirmation email never renders, and produce a ship/no-ship recommendation for the CTO. ## Quick Start Ask the agent to run the QA-UX drive against your sprint directory, for example: run /qa-ux docs/sprints/sprint-12 --mode drive --url http://localhost:3000 to verify the app before sprint acceptance.

Frequently Asked Questions about qa-ux

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run a QA acceptance test on a live web app with an AI agent?▼

Run the drive mode against your sprint directory with the app URL, for example /qa-ux <sprint-dir> --mode drive --url http://localhost:3000. The agent drives the browser via DOMShell MCP, asserts HARD signals, judges SOFT signals, and writes a qa-report.md with a ship/no-ship recommendation.

How do I write a QA test plan before development starts?▼

Use plan mode: /qa-ux <sprint-dir> --mode plan. It seeds qa-plan.md from a template where you declare in-scope surfaces, split acceptance signals into HARD (deterministic selectors/strings) and SOFT (semantic judgment) assertions, and name hero shots to capture.

What is the difference between HARD and SOFT assertions in QA testing?▼

HARD assertions are deterministic checks like exact strings, patterns, or data-testid selectors that can graduate to automated e2e tests. SOFT assertions are semantic judgments about UX quality that require agent evaluation rather than exact matching.

Why does the browser QA drive report BLOCKED for DOMShell?▼

The domshell_execute tool loads at session start, so if DOMShell was registered into .mcp.json after the session began, the tool is absent. Restart the session and re-run; do not substitute API or DB inspection for a browser drive.

Can I use the gemini engine for QA drive mode?▼

No. The gemini engine was deprecated in June 2026 after Google removed the gemini-CLI free tier, so it fails at authentication. The Claude engine is now the default and only supported path.

How are QA findings categorized and routed?▼

Every finding is tagged by severity (BLOCKER, MAJOR, MINOR, NOTE) and fault-domain (client, integration, server, data-quality, ux, unverified) with concrete evidence. Data-quality findings route to VP-DS rather than being judged statistically by QA.

Related Skills

Explore More Software Engineering Skills→
manuscomanusco

resonance-ops-product

Define product scope through Working Backwards PRDs, RICE prioritization, and PMF diagnostics.

Community
Advanced
ilandahanilandahan
9

QA & Ship Phase Skill

Validates acceptance criteria and generates release notes and deployment artifacts.

Community
Advanced
oaldersoalders
4

adversarial-review

Coordinate two reviewers to identify defects and invariant violations in code.

Community
Advanced
ray-manalotoray-manaloto
1

mise-contrib-write-e2e-test

Enforce standardized conventions for writing jdx/mise end-to-end tests.

Official
Intermediate