Skip to content

Latest commit

 

History

History
90 lines (76 loc) · 3.47 KB

custom-workflow-description.md

File metadata and controls

90 lines (76 loc) · 3.47 KB

Custom Workflow Description

Overview

This custom workflow implementation demonstrates how to integrate A2A (Agent-to-Agent) protocol compliant agents in a financial compliance workflow using the Microsoft Agent Framework.

Architecture

┌─────────────────┐    ┌──────────────────┐    ┌─────────────────┐
│                 │    │                  │    │                 │
│ Financial Agent │───▶│ Workflow Engine  │───▶│ Document Agent  │
│ (Python A2A)    │    │  (Conditional    │    │ (.NET A2A)      │
│                 │    │   Routing)       │    │                 │
└─────────────────┘    └──────────────────┘    └─────────────────┘
                                │
                                ▼
                       ┌─────────────────┐
                       │                 │
                       │ Compliance      │
                       │ Agent           │
                       │ (Mock - Always  │
                       │ Returns "OK")   │
                       └─────────────────┘

Workflow Flow

1. Financial Analysis Phase

  • Input: User query about financial performance
  • Processing: Python A2A Agent analyzes sales data, revenue trends
  • Output: FinancialAnalysisResult with structured financial metrics

2. Conditional Routing

  • Logic: If needs_policy_review == true, route to document agent
  • Decision: Based on risk areas identified in financial analysis

3a. Document/Policy Review Path

  • Input: Financial analysis results transformed to policy query
  • Processing: .NET A2A Agent searches for relevant compliance policies
  • Output: PolicyLookupResult with policy requirements

3b. Compliance Report Generation

  • Input: Combined financial + policy results
  • Processing: A2A Compliance Agent generates comprehensive report
  • Output: ComplianceReport with recommendations

3c. Alternative Direct Path

  • Condition: If no policy review needed
  • Output: Direct financial analysis summary

Data Models

FinancialAnalysisResult

{
    "total_revenue": 2450000.00,
    "top_products": ["Enterprise Suite Pro", "Analytics Dashboard"],
    "revenue_growth_rate": 15.3,
    "profit_margin": 28.7,
    "needs_policy_review": true,
    "risk_areas": ["international sales", "revenue recognition"],
    "query_context": "Original user query",
    "analysis_period": "Q4 2024"
}

PolicyLookupResult

{
    "relevant_policies": ["Revenue Recognition Policy", "Sales Policy"],
    "compliance_requirements": ["SOX compliance", "GDPR"],
    "risk_mitigation_steps": ["Quarterly audits", "Documentation"],
    "policy_categories": ["Financial", "Legal"],
    "last_updated": "2024-Q3",
    "financial_context": "Context from financial analysis"
}

ComplianceReport

{
    "executive_summary": "High-level findings summary",
    "financial_highlights": {"Revenue": "$2.45M", "Growth": "15.3%"},
    "compliance_status": "needs-attention",
    "policy_alignment": ["How results align with policies"],
    "action_items": ["Recommended actions"],
    "priority_areas": ["High-priority areas"]
}