ValidTec Projects & GitHub Repositories

How the three projects work together

The progression is now decide → control → execute. Track 1 decides what should be tested, Track 2 controls how AI receives instructions and evidence, and Track 5 proves the result through deterministic regression.

Track 1Decide what to testRequirements, risks, scenarios, approval, and a traceable automation specification.

Track 2 • NewControl AI inputsPrompt policy, context packet, project memory, token budget, evaluation, and promotion gates.

Track 5Execute and proveDeterministic API regression, CI validation, reporting, cleanup, and release evidence.

Track 1 • Agentic QA

ValidTec Track 1 — Agentic QA Workflow

Feature intake → requirement analysis → risk-based test design → human approval → automation handoff → QA reporting

This Python project demonstrates a controlled agentic QA workflow with explicit human oversight, traceable JSON evidence, automated tests, and GitHub Actions validation.

What this demonstrates

Structured feature and requirement intake
Risk-based scenario generation
Explicit human approval before automation
Traceable JSON artifacts and QA evidence
Unit tests and GitHub Actions validation

Portfolio value

Use this project to explain how you apply traditional QA controls—review, traceability, approval, and repeatability—to AI-assisted workflows.

Python 3.11+Human approval gate7 scenariosUnit testsMIT License

Workflow

Step 1Feature intake

Step 2Requirement analysis

Step 3Risk-based test design

Step 4Human approval

Step 5Automation handoff

Step 6QA reporting

View Repository →Download ZIP ↓Explore Track 1

Clone the repositorygit clone https://github.com/WaikiSin2025/ValidTec-track-1-agentic-qa.git

Repository purposeDesign, review, approve, and document AI-assisted QA scenarios before deterministic implementation.

Visitor access: Anyone can view, download, or clone this public repository. Only authorized collaborators can push changes.

Track 2 • New Prompts Repo

ValidTec Track 2 — Prompt, Context & Token Discipline

Prompt baseline → system/task separation → context packet → project memory → evaluation regression → promotion or rollback

This runnable Python project turns the completed Track 2 playbook into portfolio evidence. It treats prompts, context, memory, token budgets, and evaluation rules as versioned QA assets instead of one-off chat instructions.

What this demonstrates

Stable system policy separated from task-specific instructions
Source authority, freshness rules, context selection, and token budgeting
A concise CLAUDE.md plus reusable testing skills and known-issue records
Versioned prompt registry and structured output contracts
Representative evaluation cases with critical-case regression protection
Run telemetry, human review, promotion thresholds, and rollback evidence

Project requirements

The core demo runs with Python 3.11+ and local fixture data, so a paid model API is not required to demonstrate the QA controls. Optional provider adapters can be added later without changing the project contract.

Python 3.11+Prompt disciplineContext engineeringCLAUDE.mdJSON SchemaEvaluation gateGitHub Actions

Prompt, context & evaluation control flow

Step 1Baseline known QA cases

Step 2Separate system policy and task brief

Step 3Build authoritative context packet

Step 4Load project memory and testing skill

Step 5Run evaluation regression and telemetry

Step 6Promote candidate or roll back

View Repository →Download ZIP ↓Explore Track 2 Read Track 2 Playbook →

Clone the repositorygit clone https://github.com/WaikiSin2025/ValidTec-track-2-prompts.git

Run the deterministic demo

cd ValidTec-track-2-prompts
python3 run_demo.py
python3 -m unittest discover -s tests -v

Runnable without a paid model API: The repository uses deterministic fixture outputs to demonstrate prompt/version evaluation, context controls, token-efficiency comparison, promotion thresholds, and rollback behavior. Provider adapters can be added later without changing the QA contract.

Track 5 • New Automation Repo

ValidTec Track 5 — Automation

Health check → organization creation → user invitation → negative and authorization checks → license assignment → state verification → cleanup

This portfolio-ready regression repository converts realistic production-risk scenarios into deterministic API tests that run locally and in GitHub Actions. A built-in Node.js API keeps the demonstration self-contained.

What this demonstrates

API regression design and Postman collection structure
Newman execution and native Node.js smoke testing
Positive, negative, authorization, and entitlement checks
Test-data creation, state verification, and cleanup
GitHub Actions CI with JUnit report artifacts
Release-gate and defect-to-regression thinking

Project requirements

Node.js 20+ is required. The repository includes a small local API, Postman collection, Newman runner, smoke test, CI workflow, and synthetic test data only.

Node.js 20+PostmanNewmanNative fetchGitHub ActionsMIT License

Enterprise-style onboarding regression flow

Step 1Health check

Step 2Create organization

Step 3Invite user

Step 4Reject duplicate invitation

Step 5Reject unauthorized invitation

Step 6Assign active license

Step 7Verify onboarding state

Step 8Cleanup synthetic data

View Repository →Download ZIP ↓Explore Track 5

Clone the repositorygit clone https://github.com/WaikiSin2025/ValidTec-track-5-automation.git

Run locally

npm install
npm run server

# In another terminal
npm run smoke
npm run newman

Release gate: All onboarding P0 checks must pass. Authorization and duplicate-protection failures block release, and cleanup must succeed so synthetic data does not affect later runs.

A structure that will scale beyond three repositories

Keep the home page focused on the connected portfolio story—Track 1 decides, Track 2 controls AI inputs, and Track 5 executes. Add future repositories here using the same project-card format so the home page remains a concise visual overview rather than a technical catalog.

Browse All Learning Tracks →