The base Toolkit now takes a timeout, and HTTP timeouts are wired across the tools, with support extended to more toolkits. A tool that stalls on a slow or unresponsive endpoint now fails fast instead of holding up the run, giving you predictable behavior under real-world network conditions.
Skip to content