Tasks / Rightbrain Internal

Form Data Auditor

Automates form data extraction, normalisation, and validation, ensuring consistent and accurate information. It intelligently flags illegible or invalid fields with specific reasons, providing a clear data quality assessment.

claude-haiku-4-5-20251001
15 views
December 10, 2025

Sign up for a free trial to Clone, Customise and Run

Task Logic

User Prompt Template

CORE INSTRUCTION

Instructions sent to the model. Dynamic variables {brackets} are replaced at runtime.

Goal:
Extract all field values from a form document, normalise data formats, validate completeness, assign a completeness score, and flag any fields that fail validation or are illegible.

Input Parameters:
- {form_document}: The form document from which field values will be extracted and validated

Processing Steps:
1. Extract all field values from the form document, identifying field names and their corresponding values
2. Normalise extracted data into standard formats (e.g., dates to YYYY-MM-DD, phone numbers to international format, text to title case where appropriate)
3. Assess the legibility of each field, identifying any fields that are unclear, partially obscured, or unreadable
4. Validate each field against expected data types and formats, flagging any that fail validation rules
5. Compile a list of all flagged fields with specific reasons for validation failures or illegibility issues

Output Guidance:
Present the results in a structured format containing:
- Extracted Fields: A list of all field names with their extracted values
- Normalised Data: The same fields with values converted to standard formats
- Flagged Fields: A detailed list of any fields that failed validation or are illegible, including:
  - Field name
  - Issue type (validation failure, illegible, missing, incorrect format)
  - Specific description of the problem
  - Original value (if extractable)

Critical Rules:
- Extract all fields present in the document, even if they appear empty or illegible
- Apply consistent normalisation rules across all similar data types
- Flag fields with specific, actionable descriptions rather than generic error messages
- If a field is partially legible, extract what is readable and flag the uncertainty

Interface Definition

Inputs
form_document
Document Scanner
filename: form
Output Schema

Try this AI task with your own data

Free trial - no credit card required