Claude Code 2.1.261 Adds `/skill-doctor` to Find Skills That Go Unused

Anthropic released Claude Code 2.1.261 on September 4, 2026¹.

• /skill-doctor: shows which of the loaded skills go unused and what each one costs in context, so they can be pruned¹
• Auto mode change: a link that packs content into a public diagram renderer’s URL is now treated as an upload to that site, and is no longer auto-approved unless the user asked for it¹
• Output size settings: bashOutputMaxChars and taskOutputMaxChars raise how much command and background-task output Claude receives inline before it is saved to a file, up to 128K characters¹
• Key bindings: the prompt’s word-editing keys now match Bash — Ctrl+W deletes back to whitespace, Alt+F and Alt+D stop at word end, and punctuation separates words. keybindingFlavor no longer has any effect¹

Also in this release: --append-subagent-system-prompt-file, which reads a subagent’s system prompt from a file for prompts too large to pass on the command line; an “Organization policy” line in /status and claude doctor explaining why an organization’s policy could not be loaded; and a dangerous-rm safety prompt that now also catches rm -rf on positional parameters and inside double-quoted sh -c scripts¹. The auto mode change follows the “Containment Escape” rule added in 2.1.257 on September 1, continuing to narrow down where an agent may send data outside without asking. In environments where agents run commands automatically, changes to that line affect when approval prompts appear.

Sources

1. Claude Code v2.1.261 release notes - GitHub release notes (September 4, 2026)