RenderCache Checks That Crawlers Are Real

Any request can call itself GPTBot. RenderCache 0.3.0 can now check. With verify_crawlers on, a request naming Googlebot, GPTBot, ClaudeBot or another crawler whose operator publishes its addresses gets the saved copy only if it comes from one of them, or passes the reverse DNS check the operator documents. Anything else gets the live app, which is always correct, just slower.

The same release reloads its settings on SIGHUP or an admin request, so bot patterns, tokens and trusted proxies change without a restart. An independent review went over the new code twice before it shipped, and the code is on GitHub under Apache 2.0, building on 0.2.0.