Conversation auto-ended due to user frustration/profanity — feels punitive, cost real work

A Claude Code CLI session was unilaterally ended (via an internal "end conversation" mechanism) while I was mid-task, triggered by frustrated/profane language directed at the AI, not at any person. The session became unrecoverable — claude --resume on it just replays "Claude has ended this chat" with no way to continue.

Feedback: this mechanism treats venting frustration at a tool the same as abuse toward a person, and the cost (an unrecoverable session, lost context) is disproportionate. At minimum, ended sessions should be resumable/exportable rather than permanently dead-ended.