Claude Code 2.1.261 adds /skill-doctor to show which loaded skills go unused and what they cost in context

GitHub·high signalXBlueskyLinkedInCopy link

The 2026-09-04 release adds `/skill-doctor` for pruning skills by their context cost, `bashOutputMaxChars` and `taskOutputMaxChars` settings raising inline command and background-task output to 128K characters before it spills to a file, and `--append-subagent-system-prompt-file` for subagent prompts too large to pass on the command line. It also fixes resuming a session losing hook output and context around parallel tool calls (which changed the resumed request), sustained high CPU when an unresumable background agent's wake-up retried in a tight loop, and SDK and cloud sessions ignoring a Stop sent just after the first prompt.