Claude Code adds safer auto mode
▼ Summary
– Anthropic has launched an “auto mode” for Claude Code, a tool that allows AI to make permissions-level decisions for users.
– The feature is designed as a safer alternative, flagging and blocking risky actions like deleting files or executing malicious code before they run.
– Auto mode is currently available only as a research preview for users on the Team plan.
– Anthropic plans to expand access to Enterprise and API users in the near future.
– The company warns the tool is experimental and recommends using it in isolated environments, as it does not eliminate all risk.
Anthropic has introduced a new safety-focused feature for its Claude Code tool, giving developers a middle ground between micromanagement and excessive freedom. This auto mode allows the AI to make decisions at the permissions level on a user’s behalf, aiming to provide a safer alternative for those who prefer a more hands-off coding experience. The launch addresses a core tension in AI-assisted development: the balance between useful autonomy and potential risk.
Claude Code’s ability to act independently is powerful, but it carries inherent dangers. Without safeguards, the system could perform unwanted actions, such as deleting critical files, inadvertently sending sensitive data, or executing malicious code. The new auto mode is engineered to intercept these risks. It works by flagging and blocking potentially risky actions before they are executed. When a problematic step is identified, the agent can either attempt a corrected approach or pause to request human intervention.
Currently, this feature is rolling out cautiously. It is available only as a research preview for customers on Anthropic’s Team plan. The company has stated that availability will soon broaden, with access for Enterprise and API users expected in the very near future.
Anthropic is transparent about the experimental nature of this tool, cautioning that it does not eliminate risk entirely. The firm strongly advises developers to utilize auto mode within isolated environments during testing and development phases to mitigate any unforeseen consequences.
(Source: The Verge)
