Lightfeed open-sources TypeScript library for LLM data extraction

This library targets a common pain point in LLM-based web scraping: brittle JSON outputs that fail validation due to minor hallucinations or formatting errors.

* **Resilience over perfection:** The ability to salvage partial valid data from nested arrays or optional fields is a significant pragmatic improvement for production scraping workloads.

* **End-to-end focus:** By handling headless browser automation, content sanitization, and LLM extraction in one package, it reduces the boilerplate needed to set up reliable scraping pipelines.

* **Ecosystem flexibility:** Compatibility with LangChain ensures developers aren't locked into a single provider and can swap between local (Ollama) and hosted models.