Seven AI Skills Every QA Needs
Seven bullet-point tutorials with fifteen diagrams: NotebookLM interview prep, prompt vs skill vs agent, RAG basics, Playwright CLI, LLM evaluation, Cursor, and your own skill directory.
Start the tutorials →Every AI-for-QA guide in one place. Learn the concepts (MCP, RAG, agents), pick up a tool (LangChain, n8n, LangFlow, CrewAI), then build agents you can run against real suites.
Six modules from AI fundamentals to a shipped QA copilot. Live cohort, real projects, mentorship.
Start here. The ideas every AI-powered tester needs, each with a playable diagram.
Seven bullet-point tutorials with fifteen diagrams: NotebookLM interview prep, prompt vs skill vs agent, RAG basics, Playwright CLI, LLM evaluation, Cursor, and your own skill directory.
Start the tutorials →Anthropic's free AI Fluency course in one page: Delegation, Description, Discernment and Diligence in Anthropic's own words, the three modes of AI interaction, the certificate trap between its two official homes, and the framework applied to a QA workflow.
Start studying →The same 4D framework with the jargon taken out: a simple analogy for each D and each mode, how each one usually goes wrong, and the QA version.
Read the guide →Every key concept from the official course on one page, organised for revision: the prompt formula, Projects, Artifacts, Skills, Connectors, Research mode, and the cheatsheet.
Open the study guide →Work through Anthropic's free Claude 101 course module by module and earn the certificate: all five modules explained, projects, artifacts, skills, connectors and research, plus a fifteen-question self-test with answers.
Start studying →Anthropic free Claude Code course on one page, in its own running order, plus ten self-test questions with answers covering everything the quiz can draw on.
Read the guide →Anthropic free Introduction to agent skills course on one page, with seven hand-drawn diagrams, eight self-test questions, and every place the course and the current docs now disagree.
Read the guide →Anthropic free Introduction to MCP course on one page, with seven hand-drawn diagrams, eight self-test questions, and the SDK rename that stops the course code running today.
Read the guide →A chatbot and a RAG pipeline graded by a second language model: the five step loop, twenty five metrics grouped by the failure they catch, a twenty seven prompt attack library, and the four ways the numbers lie.
Open the teardown →Every project the 3x batch shipped: ten winners, thirty three projects with screenshots and live demos you can click through, and all forty one entries.
See the projects →Get your machine ready: local LLM, Groq key, VS Code and Copilot, Antigravity, GitHub, LangFlow, n8n, Command Code, OpenCode.
Set up my machine →The project track for the batch: twenty build-it-yourself projects that take you from a first prompt to shipped QA agents, with the docs for each.
Browse the projects →How the Model Context Protocol lets an agent drive Playwright, read your repo, and file bugs.
Read the guide →The spec revision read as a diff: sessions out, self-contained requests in, MRTR, ttlMs caching, the deprecation clock, and what it breaks in your tests.
Read the briefing →Ground AI test generation in your own specs and past bugs, so every case traces to a real requirement.
Read the guide →The reason-act-observe loop, agent frameworks, and self-healing tests, step by step.
Read the guide →Why an LLM only predicts text, while an agent acts, remembers, and verifies.
Read the guide →Prompt, vibe-coded prompt, skill file, or agent: what each is and when to reach for it.
Read the guide →Three titles, three different jobs: who trains models, who ships on foundation models, who designs the networks, and the litmus questions that decode any posting.
Read the guide →One sentence walked through the whole machine: tokens, embeddings, attention, logits, softmax, decoding, and why probable is not the same as true.
Read the guide →Three deep pillars: build chains and agents, add state and cycles, then trace and gate them. LangChain v1 accurate.
Compose testable LLM chains and agents with LCEL, tools, and RAG to generate tests and triage bugs.
Read the guide →Stateful agent graphs with cycles for retry-until-green and self-healing tests, plus multi-agent crews.
Read the guide →Trace agent runs, build datasets and evaluators, and gate CI so LLM tests stop silently regressing.
Read the guide →Two low-code pillars: build an agent on a canvas, then wire it into CI. Each with a playable diagram and a six-stage roadmap.
Three pillars for testing LLM output: pytest-native metrics, config-driven prompt matrices, and LLM-as-judge functions. Each with a playable diagram and a six-stage roadmap.
Unit-test LLM outputs in pytest with G-Eval and RAG metrics like faithfulness, then gate CI.
Read the guide →Declare prompts, providers, and assertions in one YAML file, run the matrix, and compare models side by side.
Read the guide →LLM-as-judge evals with prebuilt and custom rubrics, wired into pytest and LangSmith.
Read the guide →Build with a specific stack: orchestration, workflows, and visual agent builders.
Zero to three working QA agents: Bug Triage, RCA, and Test Designer.
Read the guide →Install n8n on local, Docker, and cloud, then wire up AI workflow nodes.
Read the guide →Three production QA agents in LangFlow and Groq: Bug Triage, Flaky Analyzer, API Contract Validator.
Read the guide →A four-agent CrewAI pipeline that turns a Jira ticket into test plans, cases, and Playwright scripts.
Read the guide →The best Claude Code plugins for SDETs: Superpowers, Caveman, Frontend Design, and more.
Read the guide →Build an API contract validation agent in LangFlow, step by step.
Read the guide →Run parallel AI coding agents side by side with cmux.
Read the guide →Hands-on projects. Copy the skill, run it, ship the report.
Triage a bug end to end: severity P0-P4, root cause, tests, and a clean HTML report.
Read the guide →Score every test for flakiness across repeat runs, on Playwright or Selenium.
Read the guide →A Copilot skill that turns a Jira story into PDF and XLSX test plans, plus a 36-skill STLC suite.
Read the guide →The Skills Masterclass companion: why a prompt has no leverage, the two shapes of a skill, where the folder lives per tool, the human review gate, guardrails, and how to validate a skill you did not write.
Read the companion →Build a SKILL.md that makes Copilot review Playwright diffs against your QA policy, and verify locators live through Playwright MCP. Works in agent mode and on PRs.
Read the guide →The capstone: a multi-source RAG copilot built over your own QA knowledge.
Read the guide →Long-form deep dives. Each one ships working code, hand-drawn diagrams, and something you can download and run.
Build agent skills that actually work: SKILL.md anatomy, progressive disclosure, and a 36-skill QA suite to download.
Read the guide →An always-on QA operator on a cheap VPS: scheduled triage, zero-token watchdogs, and coding agents dispatched from your phone.
Read the guide →The step-by-step companion: every Hermes module as a full lesson with commands and a drill.
Read the guide →Score AI output like a tester: LLM-as-judge, the metrics that matter, and a pytest-native eval framework for chatbots and RAG.
Read the guide →The terminal agent that learns your coding taste: install, VS Code sync, and the Windows setup fixes.
Read the guide →The dollar-a-month Claude Code alternative, with receipts: the CLI, the slash command catalogue, custom commands and skills, and how to make the credits go furthest.
Read the guide →Cut about 75% of your agent's output tokens with full accuracy, across Claude Code, Copilot, Codex, and OpenCode.
Read the guide →Five moves to extract a frontier model's judgment into permanent assets before it moves to pay-per-token.
Read the guide →One masterclass per assistant. Same QA jobs, each tool's own rules, skills, and agent modes.
Slash commands, CLAUDE.md, Skills, Subagents, Hooks, MCP, and Playwright MCP for agentic testing.
Read the guide →AGENTS.md, Skills, Subagents, Hooks, MCP, model routing, and AI code review for QA workflows.
Read the guide →Copilot, custom skills, and the Jira MCP to generate test plans, cases, and bug reports across the STLC.
Read the guide →Custom instructions, context variables, slash commands, agent mode, Playwright MCP, the coding agent, and the CLI.
Read the guide →Privacy-aware setup, project rules, bounded agent modes, Playwright assets, MCP, CLI review, and Bugbot.
Read the guide →Twelve lessons: rules, context, test design, Playwright automation, deterministic verification, MCP, and Bugbot.
Read the guide →The customization reference: project rules, skills, subagents, hooks, run modes and permissions, MCP, the headless CLI in CI, and Bugbot on review duty.
Read the guide →Steering, EARS requirements, feature and bugfix specs, correctness properties, hooks, Playwright MCP, and Powers.
Read the guide →Twelve lessons: specs, steering, EARS, traceability, property-based tests, hooks, MCP, and permissions.
Read the guide →Three ways to put agents on a real browser: the MCP server, the agent loop, and the agent-friendly CLI.
Isolated retries, virtual passkeys, the new component model, WebP snapshots, and the MCP server bundled in, with a one-week adoption plan.
Read the guide →Run the Playwright MCP server and drive a real browser with natural language over the Model Context Protocol.
Read the guide →The perceive, reason, act, verify loop: generate tests with AI, self-heal flaky locators, and auto-triage failures.
Read the guide →The token-efficient, agent-friendly browser command line: install, ref-based snapshots, sessions, and skills.
Read the guide →The 2-day workshop on a fully open-source stack: CLAUDE.md orchestrator, RULES.md constitution, agent skills, and a finale where BrowserBash plus a local Ollama model writes, orchestrates, and runs the automation.
Open the workshop guide →Process and reliability: spec-driven methods, verification loops, flake control, and retrieval done properly.
V6 setup, built-in QA versus Test Architect Enterprise, P0 to P3 risk design, ATDD, NFR evidence, and traceability.
Read the guide →Twelve lessons: lifecycle artifacts, TEA workflows, risk-based test design, frameworks, CI, ATDD, and traceability.
Read the guide →Bounded makers, deterministic verification, immutable oracles, evidence manifests, and Playwright repair loops.
Read the guide →All eighteen lessons: contracts, verifiers, oracles, evidence manifests, repair loops, STLC, Selenium, and AI eval.
Read the guide →Source-audited: real CLI commands, reporter contracts, scoring behaviour, the local dashboard, CI gates, and limits.
Read the guide →Twelve lessons: flake taxonomy, JUnit and JSON reports, scoring, diagnosis, dashboard, and CI guardrails.
Read the guide →A local Langflow flow with BGE-M3, Nomic, two Chroma collections, reranking, grounded answers, and tuning.
Read the guide →