Limitations
Kestrel Search is designed for lightweight public-web retrieval, not browser automation or a complete search API replacement.Provider HTML can change
Kestrel parses public HTML search pages. Provider markup, throttling, bot protection, or regional behavior can change without notice and may temporarily reduce results.No persistent cache
Search responses and fetched page bodies are not persistently cached. Repeating a command contacts the configured providers and pages again.PDF content is not fetched
PDF URLs may appear as search results, but Kestrel does not download or extract their contents.Text-oriented content only
Page fetching accepts HTML, XHTML, and plain text. Images, video, binary files, and other unsupported content types are skipped.JavaScript-rendered pages
Kestrel downloads HTTP responses but does not execute page JavaScript. Content that appears only after client-side rendering may be missing.Extraction is heuristic
Main-content extraction removes common page chrome and selects likely content containers. Unusual layouts may produce incomplete text or retain some noise.Search filters differ by provider
DuckDuckGo and Yahoo receive Kestrel’s recency filter. Bing currently ignores--time-filter. Region codes are also translated to each provider’s supported
parameter format.