Better Models, Worse Tools: Claude tool regression
Armin Ronacher's controlled tests show Claude Opus 4.8 and Sonnet 5 produce ~20% malformed tool calls against Pi's edit tool; older models are clean.
Armin Ronacher's controlled tests show Claude Opus 4.8 and Sonnet 5 produce ~20% malformed tool calls against Pi's edit tool; older models are clean.