Open-source tool bulk-pdp-generator: Automatically generate e-commerce product detail pages with Claude

bulk-pdp-generator is an open-source Python-based two-stage LLM pipeline tool that can batch generate SEO-optimized product description HTML (including structured data markup) from Shopify product data. It solves the content gap problem for e-commerce websites with large-scale SKUs, and is suitable for Shopify brands while being adaptable to other platforms.

Most e-commerce websites face content issues: empty product page descriptions, placeholder text, content only present in images (invisible to crawlers), or lack of structured data (e.g., aggregateRating). These problems lead to low search engine rankings, especially when SKU scale reaches hundreds or thousands, becoming a data pipeline challenge.

Research Phase: Parse product tags/titles, call Claude to generate a 13-line attribute table (including product type, ingredient efficacy, etc., with unverified content marked [UNVERIFIED]); Writing Phase: Generate HTML containing core paragraphs, usage steps, and JSON-LD structured data based on the attribute table. Supports bundle processing: parse pipe-separated titles to identify components, cross-reference product catalogs to ensure description accuracy.

The product pages of Plum Goodness, an Indian D2C skincare brand, are visually appealing but lack crawlable text content and structured data. This tool uses structured data from its Shopify tags (e.g., SKINTYPE_All skin types) to generate SEO-compliant descriptions, solving the content display problem.

Tech Stack: Python 3.10+, Claude Sonnet (Anthropic API), input Shopify products.json, output JSON/Excel, etc.; Features: Empty product filtering, single product mode, review data integration, Shopify-ready output, Excel export, dry run; Cost: Approximately $36-37 for Claude Sonnet to process 1000 SKUs.

The tool is based on Plum Goodness's setup and requires adaptation for: tag taxonomy (different store tag rules), attribute tables (adapted for skincare, needs adjustment for other categories), structured data types (e.g., Book/SoftwareApplication), review sources (tools other than JudgeMe), output formats (column remapping needed for WooCommerce/Magento). Adaptation guidelines can be found in docs/ADAPTING.md.

1. Installation: git clone https://github.com/bobbymathew-lab/bulk-pdp-generator.git → pip install -r requirements.txt; 2. API Setup: export ANTHROPIC_API_KEY="your-key"; 3. Dry Run: python plum_pdp_generator.py --input examples/sample_products.json --filter-empty --dry-run; 4. Generate Descriptions: python plum_pdp_generator.py --input products.json --output results.json --filter-empty; 5. Export to Excel: python results_to_excel.py --input results.json --output review.xlsx.

This tool demonstrates the practical value of LLMs in e-commerce content generation: large-scale processing (thousands of SKUs compressed to a few hours), quality assurance (anti-hallucination instructions), SEO optimization (structured data), and adaptability (provides guidelines for expansion). It helps e-commerce brands reduce operational costs and improve search visibility and conversion rates.