AI Agents Keep Failing the Same Ways: Poor Documentation, Hallucinations, and the "Interesting" Non-Fix

Developers continue documenting a frustrating pattern: AI agents that confidently claim to fix bugs they haven't fixed, hallucinate when documentation is poor, and lack the self-awareness to flag their own uncertainty.