News on October 29: According to tech media The Register, Anthropic recently quietly added a brand-new environment variable in the released Claude Code v2.1.311. By enabling this setting, users can allow AI agents to perform high-risk operations such as deleting files without seeking user approval each time.
The full name of this brand-new environment variable is CLAUDE_CODE_EVEN_MORE_DANGEROUSLY_SKIP_PERMISSIONS, which literally translates as "even more dangerously skip permission checks".
Anthropic technologist Daisy Hollman explained the origin of this feature in a subsequent response. Although the --dangerously-skip-permissions (dangerously skip permission checks) option built into previous versions sounded extremely powerful, it actually did not completely allow all high-risk operations; the system's underlying layer would still forcibly block some obviously dangerous commands, such as rm -rf ~, which recursively and forcefully deletes the contents of the user's home directory. The newly added extreme environment variable was created specifically to meet the extremely high autonomy needs of some core developers, allowing them to completely shut off this final line of defense.
Regarding the questions raised about the variable's overly blunt and slightly tongue-in-cheek naming, as well as its absence from the official changelog of the corresponding version, Hollman subsequently publicly confirmed on social media: "This is not a joke."
However, it should be noted that this feature is not enabled by default for all users; it must be actively configured by developers in their environment to take effect, and the relevant security policies and restrictions previously imposed by organization administrators will also continue to apply. For senior developers seeking maximum efficiency who want to grant AI agents full autonomous operating permissions, this update undoubtedly greatly reduces the tedious pop-up confirmation process; but from a system security perspective, how to find a balance between complete delegation and preventing misoperations will also become a new issue that developers need to weigh carefully in actual use.