jarencudilla

k6 on Windows Without WSL or Admin Rights

k6 ships as a standalone binary. No installer, no package manager, no elevated permissions required.

Setup

Extract k6.exe to a folder your user account controls, then call it from Git Bash:

~/k6/k6.exe run your-test.js

jarencudilla

pw-script Intake Protocol

Before generating a Playwright script, establish all of the following. Missing answers produce flaky or incorrect automation.

Required before generation

A. Surface — exact feature, page, or component. URL or route if available.

B. Auth and role — which user role does this reproduce under. Does a storageState fixture exist for this role?

jarencudilla

BYOK AI Tool Stack — Reference

Model: Bring your own key. API calls go directly from browser to provider. No backend, no server-side key storage, no platform markup.

Cost per use: Fractions of a cent on current Claude/OpenAI/Gemini plans for most tool operations. For heavy daily use, route lighter tasks to a local LLM (Ollama) to reduce frontier model spend.

Stack:

• Plain HTML + vanilla JS
• No framework, no build step
• GitHub Pages for deployment

jarencudilla

AI for QA: Prompts That Generate Real Test Value

Published on QAJourney.net

AI isn't a magic test generator—it mirrors whatever context it's given. With requirements, acceptance criteria, and supervision, AI expands coverage fast. Without it, you get fluff.

The Real Problem It Solves

• AI generates shallow/happy-path cases by default
• Lacks UX judgement + business logic awareness
• Blind automation amplifies mistakes

jarencudilla

Guerrilla QA: Shift-Left Testing for the Real World

Published on QAJourney.net

Not every team has devices, tools, or time. Guerrilla QA succeeds by prioritizing what breaks first, not what looks thorough on paper. Fewer documents, faster feedback, more real coverage.

The Real Problem It Solves

• Testers wait for resources instead of executing
• Enterprise-style process slows small teams
• Feature velocity outpaces documentation

jarencudilla

QA Methodology vs Test Cases: Coverage Is the Real Failure Point

Published on QAJourney.net

Teams often blame Agile/Waterfall/Hybrid models when bugs still leak into production—but the real issue is shallow test case design. Weak suites create false confidence, duplicated effort, and regression cycles that don't actually protect releases.

The Real Problem It Solves

• Methodology changes don’t fix missing test depth
• Suites pass while production still breaks
• Duplicate or shallow cases inflate coverage falsely
• Regression becomes ritual instead of protection

jarencudilla

Annotation Tools: The Fix for Sloppy Bug Reporting

Published on QAJourney.net

Most teams lose hours explaining screenshots instead of fixing bugs. Annotation tools erase that drag by giving developers exact visual context, browser details, and pinpointed failure zones. Less guessing, fewer Slack threads, and zero “where is this?” follow-ups. It’s not about pretty screenshots — it’s about eliminating ambiguity from defect communication.

The Real Problem They Solve

• Screenshots that lack context or environment data
• Missing reproduction clarity leading to back-and-forth messages
• Bugs misunderstood because of unclear indicators

jarencudilla

QA Villain Perception: How Teams Misread the Work

Published on QAJourney.net

Teams love the idea of “velocity” until QA shows the real cost of rushing work upstream. That’s when the villain mask gets thrown at the testers — not because QA caused delays, but because QA exposes where the delay was already created. The frustration isn’t personal. It’s structural: uneven workload distribution, undefined acceptance criteria, and tasks piling at the tail end of the sprint.

Where Perception Breaks

• QA is pulled in only after development runs late
• Requirements shift while testing is already in progress
• Defects get treated as “QA blocking the sprint”

jarencudilla

Teaching LLMs Data Evaluation via Prompt Engineering

Premise:
LLMs don’t “understand” data — they approximate patterns. We taught them how to grade datasets instead of hallucinating conclusions.

Why it matters:
Every AI team hits the “data blindness” wall. Prompt engineering can simulate reasoning — if structured like QA logic.

Ignore:
Over-complicated “autonomous agent” frameworks.

jarencudilla

Building the EAI Anti-BS Filter: A WordPress Spam Plugin

Goal:
Catch AI-generated spam and SEO sludge before moderation fatigue sets in.

Build:
A plugin that scores comment intent and entropy — detects hollow AI fluff in seconds. If it sounds “too polite,” it’s filtered.

Why it matters:
AI spam floods WordPress faster than basic filters catch it. This one spots tone patterns, not just keywords.