Long-Running AI Agent

An AI agent capable of running continuously for days or weeks, autonomously advancing complex tasks over extended periods. It possesses self-correction and iteration capabilities, with core challenges centering on cost management and reliable self-evaluation mechanisms.