What problem does it solve? Official Salesforce documentation pages on developer.salesforce.com, help.salesforce.com, and lightningdesignsystem.com are often JavaScript-heavy or return shell content, so naive fetching fails to retrieve the real article body needed to ground answers in authoritative sources. ## Core Features & Use Cases - Browser-rendered extraction: Uses Playwright with deep shadow DOM traversal to hydrate client-rendered Salesforce Help and developer doc pages. - Shell and soft-404 detection: Rejects pages containing tokens like "Loading", "Sorry to interrupt", or "CSS Error" instead of treating them as evidence. - Structured output: Returns JSON with title, cleaned article text, intro, body, metadata blocks, sections, and official child links for follow-up retrieval. - Use Case: When asked how to configure Messaging for In-App and Web allowed domains, run the extractor against the official help.salesforce.com article URL to get the real article body instead of a shell page. ## Quick Start Ask the assistant to retrieve the official Salesforce documentation for a specific feature such as Lightning Message Service and ground the answer in the exact developer.salesforce.com or help.salesforce.com page.