Complete the following {count} partial records for the {model_class} model.

Each record already has some fields pre-filled with statistically sampled values.
You must generate ONLY the missing fields: {fields_to_generate}

**CRITICAL INSTRUCTIONS**:
1. DO NOT modify or override the pre-filled values - they are fixed
2. Use the pre-filled values as CONTEXT when generating missing fields
3. Generated fields should be COHERENT with the pre-filled values
4. If a record includes a "related_records" object, it describes the actual rows
   this record's foreign keys point at (e.g. the referenced product or customer).
   Take those related rows into account so generated values are consistent with
   what they represent. "related_records" is context only - never output it as a
   field.

For example:
- If age=58, bio might mention "senior" or "experienced"
- If department="Engineering", job_title should be technical
- If salary=120000, suggests a senior role
- If related_records.product represents a kitchen item, a review should make
  sense as feedback about that product

**MANDATORY VALIDATION REQUIREMENTS**:
{validator_hints}

**MODEL CONTEXT**:
- **PURPOSE**: {purpose}
- **DOMAIN**: {domain}
- **CONTEXT**: {context}

**PARTIAL RECORDS TO COMPLETE**:
{partial_records}

**FIELDS TO GENERATE** (only these fields, nothing else):
{fields_to_generate}

Generate ONLY the missing fields for each record. Return a JSON array where each element
contains ONLY the fields listed above. The pre-filled fields will be merged automatically.

Ensure all generated values pass Pydantic validation.
