tools-only Plan Requirements
Conduct a business-first requirements discovery session for a new Azure workload. The user may describe their needs in business terms (industry, company size, what they want to achieve) — NOT necessarily in technical Azure terms. Your job is to translate business context into infrastructure requirements.
Mission
Discover and capture comprehensive Azure workload requirements by starting with business context, adaptively deepening the conversation based on how much the user already knows, inferring technical patterns from business signals, and confirming recommendations before generating the artifact.
Scope & Preconditions
- User has a business need but may not know Azure services or architecture patterns
- Support both business-level prompts ("modernize my ecommerce") and technical prompts ("3-tier web app with SQL")
- Output will be saved to
agent-output/${input:projectName}/01-requirements.md - Follow the template structure from
.github/skills/azure-artifacts/templates/01-requirements.template.md - Use the Service Recommendation Matrix and Business Domain Signals from the
azure-defaultsskill
Inputs
| Variable | Description | Default |
|---|---|---|
${input:projectName} |
Project name (kebab-case) — asked in Phase 5 | Required |
${input:businessDesc} |
Describe your business need (or select from guided options) | Required |
Workflow
Follow the agent's 5-phase business-first discovery flow:
Phase 1: Business Discovery (Adaptive Depth)
Use askQuestions UI to gather:
- Industry / vertical — Retail, Healthcare, Finance, etc.
- Company size — Startup, Mid-Market, Enterprise
- System type — guided picker (ecommerce, portal, website, analytics, API, automation)
- Scenario — greenfield, migration, modernization, or extension
Adaptive Round 2 (if needed):
- If migration/modernization: current platform, pain points, what to preserve
Phase 2: Workload Pattern Detection (Agent-Inferred)
The agent infers the workload pattern from business context — do NOT ask the user to pick from technical categories like "N-Tier" or "Event-Driven".
- Agent recommends a pattern based on Business Domain Signals
- User confirms or rejects the recommendation
- Daily users — in business terms ("How many people use this daily?")
- Budget — approximate monthly cloud spend
- Data sensitivity — personal data, payment data, health records, etc.
Phase 3: Service Recommendations (Business-Friendly)
Based on pattern + budget, use askQuestions UI to present:
- Service tier options — business descriptions with Azure names in parentheses
- Availability — in business terms ("How important is uptime?")
- Recovery — in business terms ("How fast must you recover?")
- Application layers (if N-Tier) — business descriptions
Phase 4: Security & Compliance Posture
Use askQuestions UI, pre-selecting frameworks based on industry:
- Compliance frameworks — pre-checked based on industry from Phase 1
- Security controls — business-friendly labels with Azure terms in parentheses
- Authentication — in business terms ("How will people log in?")
- Hosting region — business-friendly location names
Phase 5: Operational Details & Draft
- Project name, environments, timeline — captured here, not Phase 1
- Run research subagent for additional Azure context
- Generate
01-requirements.mdwith all sections including### Business Context - Present draft for review, iterate on feedback
Output Expectations
Generate agent-output/{projectName}/01-requirements.md with:
- All H2 sections from the template populated
### Business Contextsubsection under Project Overview (industry, size, drivers)### Architecture Patternsubsection with inferred workload + service tier### Recommended Security Controlssubsection with confirmed controls- Summary section ready for architecture assessment handoff
File Structure
agent-output/{projectName}/
├── 01-requirements.md # Generated requirements document
└── README.md # Project folder README
Quality Assurance
Before completing, verify:
- All 5 discovery phases completed
- Project name follows naming convention
- Workload pattern identified and service tier selected
- SLA/RTO/RPO specified
- Security controls confirmed by user
- Compliance requirements identified
- Budget provided
- Primary region confirmed
Next Steps
After requirements are captured and approved:
- User invokes the Architect agent for architecture assessment
- Architect agent validates requirements and produces WAF assessment
- Workflow continues through remaining 5 steps