Try it here: https://https://github.com/Dmitriusan/GptMonkey
["Checkout the project, create application-local.properties with secrets like\n\nhawk.db.url=jdbc url\nhawk.db.username=hawk\nhawk.db.password=\nopenai.token=sk-****\n\nperform generateAllApiClients task at Gradle, and then build&run the project. It currently supports only looking for inline skate wheels (hardcoded query and heuristics/LLM prompting), and outputs findings & highlighted items into the DB (results of multiple runs are merged). The pipeline itself and class hierarchy is generic, so other categories/queries/extractors/matchers can be added by writing the code"]
About
This buying bot for eBay will be optimized for a high-volume automatic search, and I see the business model as B2B SaaS. eBay is a great marketplace to apply LLM-based tools for scanning because listings are fuzzy and hard to process in an automatic manner. At the same time, prices here are often well below the manufacturer's suggested retail price (MSRP) or recommended retail price (RRP) (especially on used/refurbished items), and some items are just unique. So the idea is to apply various heuristics (statically hardcoded and AI-based), and extract enough cherry-picks to make this "lot mining" commercially sustainable.