brand_truth_create
Capture brand truths from natural-language text. Submit prose, and the extraction pipeline stores atomic facts and reports added/skipped with reasons.
Instructions
Capture brand truths from plain natural-language text. Describe facts/observations about the brand in ordinary prose — a sentence or short paragraph (≤8000 chars). Do NOT pre-classify, pre-format, or guess the taxonomy: PostKing's LLM extraction pipeline reads the text and decides which atomic 'brand truths' to persist and which to skip (duplicates, too vague, or previously rejected facts). Returns { addedCount, skippedCount, added, skipped }: 'added' are the truths that were stored; 'skipped' lists each rejected quote with a reason. Review both — the skipped reasons tell you what didn't make it in and why.
Input Schema
| Name | Required | Description | Default |
|---|---|---|---|
| text | Yes | Free-form natural-language description of facts/observations about the brand (≤8000 chars). Write it as plain prose — the extraction pipeline handles classification. | |
| brandId | No | Brand ID (defaults to active brand) |