When we introduced
raghilda
in April, the package handled the core RAG workflow: read documents,
chunk them, embed them into a store, and retrieve relevant chunks at
query time. raghilda v0.2 broadens that scope considerably. The
release adds a structured crawl and ingest API that replaces the manual
read-and-upsert loop with a pipeline built around caching, concurrency,
and composable crawlers (including a CloudflareCrawler that can index
JavaScript-rendered sites without running a local headless browser). It
also ships a PostgreSQL store backend and NVIDIA NIM embedding support.
This is a companion discussion topic for the original entry at https://opensource.posit.co/blog/2026-07-01_raghilda-0-2-0/