In a move anticipated by its user base, Anthropic has updated its Claude Code AI assistant with an 'auto' mode, allowing the tool to manage its own file and system permissions. This shift addresses a primary complaint from experienced developers: constant approval requests that disrupted their workflow.

The feature, reported by TechRadar, does not offer blanket access. Instead, Claude Code uses a context-aware model to assess risk. It might silently read a configuration file but will still request confirmation before executing a high-stakes shell command. Developers can configure the boundaries, choosing a level of autonomy that matches their project's risk profile.

For Anthropic, this is a competitive necessity. The market for AI coding assistants is crowded, with rivals like GitHub Copilot and OpenAI's Codex prioritizing speed. A tool that feels slow or intrusive risks being abandoned. The update formalizes a practice many developers had already engineered themselves using scripts and workarounds.

The change touches a central debate in software engineering: how much independence should an AI have? Some treat these tools as advanced autocomplete, while others integrate them into core development tasks. Auto mode lets developers choose their own approach.

Anthropic, a company known for its public focus on AI safety, is making a pragmatic calculation. An overly cautious tool may push users toward less restrained alternatives. The company has built in safeguards, including audit trails for post-action review. The coming months will show if this balance of autonomy and oversight can sustain developer trust during daily use.

Source: Webpronews