Instruction file imported from gdosiadis/market-news-app-dotnet (.github/instructions/dotnet.instructions.md). Copyright stays with the author.

.NET / C# Guidelines

• Target framework is .NET 8 with nullable reference types enabled — respect ? annotations, don't silently introduce nulls without updating signatures.
• This is a top-level statements program (Program.cs); keep new orchestration logic there and business logic in Services/.
• All I/O (scraping, AI calls, email, file cache) is async/await end-to-end — never block with .Result/.Wait() on a Task.
• Playwright calls (Services/Scraper.cs) must be defensive: wrap per-element/per-site operations in try/catch so one selector or one site failing doesn't abort the whole scrape — this mirrors the existing pattern of returning partial results with a status per source rather than throwing.
• When adding a new scraped site, follow the existing SiteConfig pattern (selectors, optional FollowFirstLinkSelector for list-style pages) instead of one-off scraping logic in Program.cs.
• Prefer extending the existing cache-hash mechanism (ScrapeCache.cs/SummaryCache.cs) over adding a second, parallel caching scheme.