[charcircuit]

Models like Claude have been trained to update and reference memory for Claude Code (agent loops) independently and as a part of compacting context. Current models have been trained to keep learning after being deployed.

[ra]

yes but that's a very unsatisfactory definition of memory.