Application API
Complete reference for application processing prompt builders
All application processing functions are prompt builders — they return prompt strings for use with any LLM.
Detection
APPLICATION_CLASSIFY_PROMPT
A static prompt string for classifying whether a PDF is an insurance application form.
import { APPLICATION_CLASSIFY_PROMPT } from "@claritylabs-inc/cell";Field extraction
buildFieldExtractionPrompt()
Build a prompt to extract all fillable fields from an application PDF as structured data.
Returns: string
Field types: text, numeric, currency, date, yes_no, table, declaration
Auto-fill
buildAutoFillPrompt(fields, businessContext)
Build a prompt to match extracted fields against known business context for auto-filling.
Returns: string
Question batching
buildQuestionBatchPrompt(unfilledFields)
Build a prompt to organize unfilled fields into topic-based batches for user interaction.
Returns: string
Answer parsing
buildAnswerParsingPrompt(questions, userReply)
Build a prompt to parse free-text user replies into structured field values.
Returns: string
Email generation
buildBatchEmailGenerationPrompt(batch, context)
Build a prompt to generate an email asking users for missing application information.
Returns: string
Reply classification
buildReplyIntentClassificationPrompt(userReply, currentBatch)
Build a prompt to classify user reply intent (answering questions, asking for help, etc.).
Returns: string
Field explanation
buildFieldExplanationPrompt(field)
Build a prompt to generate a human-readable explanation of a complex insurance form field.
Returns: string
Confirmation
buildConfirmationSummaryPrompt(filledFields)
Build a prompt to generate a summary of all filled fields for user review before PDF generation.
Returns: string
PDF mapping
buildAcroFormMappingPrompt(filledFields, acroFormFields)
Build a prompt to map filled field values to AcroForm field names in a fillable PDF.
Returns: string
buildFlatPdfMappingPrompt(filledFields, pageInfo)
Build a prompt to map filled field values to percentage-based page coordinates for flat PDF overlay.
Returns: string
Lookup fill
buildLookupFillPrompt(fields, documents)
Build a prompt to fill fields by looking up values from existing insurance documents.
Returns: string