Azure DevOps PBI Analyzer
Analyze Azure DevOps Product Backlog Items (PBIs) to generate test scenarios, Gherkin specifications, Cypress/Playwright tests, and comprehensive gap analysis using AI-powered agent orchestration.
✨ Features
- 🔄 Deep Analysis: Recursively fetch work items with full hierarchy (parents, children, dependencies)
- 🧠 AI-Driven Gap Analysis: Detect missing NFRs, contradictions, edge cases, and integration gaps
- 🎭 Multi-Persona QA Analysis: Copilot Chat with 6 QA personas
- 🤖 27 AI Agents: Install expert agents for testing, architecture, debugging, and more
- 📝 Auto-Generated Tests: Gherkin, Cypress, Playwright, and Manual test cases
🚀 Quick Start
Option 1: Use the VS Code Extension
- Install Extension from VS Code Marketplace
- Analyze a PBI:
- Press
Ctrl+Shift+P
- Run:
QA the PBI
- Enter PBI URL:
https://dev.azure.com/org/project/_workitems/edit/12345
- View Analysis with deep hierarchy and AI-generated questions
Option 2: Use AI Agents Directly (No Extension Needed!)
Works with ANY AI model (ChatGPT Free, Claude, Gemini, etc.)
Install Agents:
- Press
Ctrl+Shift+P
- Run:
Azure DevOps: Install AI Agents and Workflows
- OR manually copy
.agent folder to workspace
Use with GitHub Copilot (Free or Paid):
#file:.agent/agents/qa-engineer.md
Create test cases for: [Your PBI]
OR Copy-Paste to Any AI:
- Open
.agent/agents/qa-engineer.md
- Copy entire content
- Paste into ChatGPT/Claude/Gemini
- Add your question below
📖 Full Guide: See .agent/HOW_TO_USE.md for detailed instructions
🎭 Copilot Chat Commands
Chat Participant: @azdo.pbi-analyst
| Command |
Description |
/analyze |
Full multi-persona orchestration (all QA personas) |
/critical-review |
Critical thinking analysis (assumptions, gaps, risks) |
/qa-engineer |
Test case design and coverage analysis |
/qa-architect |
Testing strategy and framework recommendations |
/manual-test |
Exploratory testing and UX focus |
/automation |
Test automation strategy and implementation |
Example:
@azdo /analyze
[Your PBI details]
🤖 Available AI Agents (27 Total)
QA Specialists:
- 🧠 Critical Thinker - Requirements analysis, assumption identification
- 🎭 QA Orchestrator - Coordinates personas, synthesizes insights
- 🧪 QA Engineer - Test case design, functional testing
- 🏗️ QA Architect - Testing strategy, framework selection
- 🔍 Manual Tester - Exploratory testing, UX validation
- ⚙️ Automation Engineer - Test automation, CI/CD integration
Development Specialists:
- 🎨 Frontend Specialist - UI/UX, React, TypeScript
- 🔧 Backend Specialist - APIs, databases, architecture
- 📱 Mobile Developer - iOS, Android, React Native
- 🛡️ Security Auditor - Vulnerability scanning, compliance
- 🐛 Debugger - Root cause analysis, troubleshooting
- 📊 Database Architect - Schema design, optimization
... and 15 more! See .agent/agents/ folder.
🛠️ Alternative Usage (No Extension Required)
For VS Code Users (GitHub Copilot)
In Copilot Chat:
#file:.agent/agents/critical-thinker.md
Analyze this PBI:
"Users can reset password securely"
For ChatGPT / Claude / Gemini Users
- Open
.agent/agents/critical-thinker.md
- Copy all content
- Paste into AI chat
- Add your question
No VS Code needed! No installation needed! Works with FREE AI models!
📋 What Gets Analyzed?
Deep Work Item Hierarchy:
- ✅ Parent chain (Epic → Feature → Grandparent)
- ✅ Children (Tasks, Bugs, Sub-tasks)
- ✅ Dependencies (Related items)
Enhanced Gap Analysis:
- 🔒 Security requirements missing?
- ⚡ Performance targets undefined?
- ♿ Accessibility not mentioned?
- 🚨 Error handling gaps?
- ✔️ Validation missing?
- 📊 Logging/monitoring absent?
- 🧪 Edge cases not covered?
- 🔗 Integration timeout/retry strategy?
Auto-Generated Artifacts:
- Test scenarios (Positive, Negative, Edge cases)
- Gherkin specifications
- Cypress test pseudocode
- Playwright test pseudocode
- Manual test cases (step-by-step)
📖 Documentation
💡 Examples
Example 1: GitHub Copilot Chat
@azdo /analyze
PBI: "Implement OAuth2 login for mobile app"
Acceptance Criteria:
- Users can log in with Google and Facebook
- Session persists for 30 days
- Token refresh happens automatically
AI orchestrates:
- Critical Thinker → Identifies PKCE compliance gap
- QA Engineer → Creates 12 test scenarios
- QA Architect → Recommends Playwright + Detox
- Automation Engineer → 80% automation target
Example 2: Using Agents with ChatGPT Free
Step 1: Copy .agent/agents/qa-engineer.md content
Step 2: Paste into ChatGPT:
[Pasted QA Engineer agent prompt]
Create test cases for:
Feature: Shopping cart checkout
- User adds items
- Applies discount code
- Completes payment
Step 3: Get expert test cases with Given/When/Then format!
🆓 Works with Free AI Models - No Quality Compromise
These agent prompts are engineered for free models. Get professional, expert-level results without paid subscriptions.
✅ GitHub Copilot Free - ⭐⭐⭐⭐⭐ Excellent
✅ ChatGPT Free (GPT-4o mini) - ⭐⭐⭐⭐⭐ Excellent
✅ Claude Free - ⭐⭐⭐⭐⭐ Excellent
✅ Gemini Free - ⭐⭐⭐⭐ Very Good
Why Quality Is Identical:
Each agent prompt includes:
- ✅ Complete frameworks (self-contained, no external knowledge needed)
- ✅ Structured templates (free models follow them perfectly)
- ✅ Concrete examples (guides output format)
- ✅ Quality checklists (built-in validation)
Result: Free models produce the same expert-level analysis as paid models. The structured prompts compensate for any capability differences.
📖 Optimization Guide: .agent/FREE_MODELS_GUIDE.md
🔧 Configuration
Settings (Ctrl+,):
Azure DevOps PBI Analyzer: Organization - Default org URL
Azure DevOps PAT:
- Required for API access
- Scopes: Work Items (Read)
- Stored securely in VS Code Secret Storage
📦 Installation
From VS Code Marketplace
- Search "Azure DevOps PBI Analyzer"
- Click Install
From .vsix File
code --install-extension azdo-pbi-analyzer-1.0.3.vsix
🤝 Contributing
See CONTRIBUTING.md
📄 License
MIT License - See LICENSE
🎯 Key Takeaway
You DON'T need the extension to use the AI agents!
The 27 expert agents work with:
- ✅ GitHub Copilot (Free or Paid)
- ✅ ChatGPT Free
- ✅ Claude Free
- ✅ Any other AI model
Just copy-paste or use #file: references in Copilot Chat!
📘 Get Started: Read .agent/HOW_TO_USE.md