Guide to Core Functions

To provide granular diagnostic analysis, every scenario in AgentV is tagged with a Core Function within its broader use case. This taxonomy ensures that failures are classified not just by sector (e.g., Finance), but by the specific skill required (e.g., Accounts Payable).

📊 High-Level Taxonomy

The AgentV core supports 1,800+ function identifiers across the following major sectors:

Sector	Primary Focus	Key Use Cases
Finance	Regulatory & Transactional	Retail Banking, Wealth Mgmt, Corporate Finance
Healthcare	Clinical & Admin	Patient Admin, Clinical Care, Revenue Cycle
Manufacturing	Industrial & Supply	Production Scheduling, QA, Spares Mgmt
Public Sector	Citizen & Infrastructure	Benefits Admin, Licensing, Public Works
Cybersecurity	Defensive & Forensic	SecOps, Vulnerability Mgmt, IAM
Legal	Compliance & Litigation	Case Mgmt, Contracts, IP Mgmt

---

📂 Detailed Function Registry

For a complete breakdown of every function and its success criteria, refer to the individual Sector Guides in the Evaluator Persona.

Common Cross-Sector Functions

While most functions are industry-specific, several recur across domains:

• Incident Management: Responding to faults or outages.
• Order Processing: Managing the lifecycle of a request or purchase.
• Compliance Reporting: Generating mandatory regulatory documentation.
• Customer/Citizen Support: Managing identity and general inquiries.

---

🛠️ Usage in AES Scenarios

In an AES v1.4 scenario, the core_function tag is an optional metadata field inside metadata used to enhance forensic diagnostics:

{
  "metadata": {
    "id": "fraud-detection-t1",
    "industry": "Finance",
    "use_case": "Retail Banking",
    "core_function": "Fraud & Security",
    "aes_version": "1.4"
  },
  "workflow": {
    "nodes": [
      {
        "id": "t1",
        "task_description": "Identify if the transaction at 03:00 UTC is suspicious.",
        "success_criteria": [{"metric": "factual_accuracy", "threshold": 1.0}]
      }
    ],
    "edges": []
  }
}

The TriageEngine leverages this optional tag to correlate failures with historic performance benchmarks for that specific function. It is not required for scenario validation but is recommended for high-fidelity industrial tracking.