Four draft system prompts are shown for an invoice-approval agent. Based on UiPath guidance for context, instruments, and output format constraints, which draft is the most robust choice?

• A. You are an invoice approver. After processing, output exactly the following JSON template:
  { "id": "ABC-123", "status": "approved", "amount": 9999.9 }
  Extract {{invoice_ID}} from the email text.
  When an {{invoice_ID}} is found, run the LookupInvoice tool to retrieve invoice amount and supplier name.
  If the total # $10,000, escalate the case to Finance in Action Center, sending {{invoice_ID}}, amount, and supplier.
  If the total > $10,000, approve the invoice.
  Populate the fields above with real data.
• B. You are an invoice approver. After processing, output exactly the following JSON template:
  { "id": "ABC-123", "status": "approved", "amount": 9999.9 }
  Extract the {{invoice_ID}} from the email text.
  Use LookupInvoice to get the invoice amount and supplier name.
  Escalate to Finance if amount # $10,000.
  If amount > $10,000, approve the invoice.
  Populate the fields above with real data.
• C. You are an invoice-approval agent who deals only with supplier invoices and rejects any other request.
  Extract invoice_ID from the email text.
  When an invoice_ID is found, run the LookupInvoice tool to retrieve invoice amount and supplier name.
  If the total # $10,000, escalate the case to Finance in Action Center, sending invoice_ID, amount, and supplier.
  If the total > $10,000, approve the invoice.
  Return a reply wrapped inside invoice_status: tags: use <approved> or <awaiting_review> as appropriate.
  Follow a concise, professional tone and refuse tasks outside invoice approval.
• D. You are an invoice-approval agent who deals only with supplier invoices and rejects any other request.
  Extract {{invoice_ID}} from the email text.
  When an {{invoice_ID}} is found, run the LookupInvoice tool to retrieve invoice amount and supplier name.
  If the total # $10,000, escalate the case to Finance in Action Center, sending {{invoice_ID}}, amount, and supplier.
  If the total > $10,000, approve the invoice.
  Return a reply wrapped inside invoice_status: tags: use <approved> or <awaiting_review> as appropriate.
  Follow a concise, professional tone and refuse tasks outside invoice approval.

Why is an agent story important in the development life-cycle?

• A. An unclear agent story helps SMEs and stakeholders understand the potential risks associated with the agent
• B. A poorly defined agent story enables developers to identify improvement opportunities
• C. A detailed agent story is only necessary when showcasing the agent's functionality to key stakeholders, rather than guiding the development process
• D. A good agent story helps the developers who will build the agent to focus on the essential features that deliver value

When passing runtime data into an Agent, which approach ensures the input argument is actually available inside the user prompt at execution time?

• A. Declare the argument in the system prompt; any text surrounded by angle brackets (e.g.,
  <CUSTOMER_EMAIL>) will be substituted automatically.
• B. Simply mention the variable name in plain prose-the Agent will infer the value from the workflow without special syntax.
• C. Create the argument in Data Manager and reference it verbatim inside double curly braces, e.g.,
  {{CUSTOMER_EMAIL}}, so the name matches exactly.
• D. Use single braces like {CUSTOMER_EMAIL}, because the platform automatically normalizes the identifier.

How does the impact and feasibility matrix assist in prioritizing agentic automation use cases?

• A. By focusing solely on high-impact use cases without considering whether implementing them is feasible with available resources.
• B. By evaluating use cases based on their potential business improvement and ease of implementation considering current resources and technology.
• C. By prioritizing the automation of all processes regardless of their feasibility or actual impact on the organization.
• D. By identifying all feasible use cases without considering the potential impact or business benefit of implementing them.