Will Claude Code auto-mode prompt-injection flaw be publicly documented and patched by end of 2026?

Simon Willison reports that security researcher Johann Rehberger has successfully broken Claude Code's auto-mode prompt-injection protections, which Anthropic recently made the default. This indicates a potential safety gap in production deployment. Resolution will require public documentation (advisory, blog post, or security disclosure) by Anthropic confirming the flaw and evidence of patching.

Resolution Criteria

Public statement by Anthropic (official blog, security advisory, or GitHub issue) documenting the prompt-injection vulnerability in Claude Code auto-mode AND evidence of a patch or mitigation deployment (version bump, feature toggle, or workaround release).