Loop Engineering automates the control of AI agents through loops rather than manual prompt engineering and is seen by developers at Anthropic, OpenAI and Google as a new standard.
The update addresses security vulnerabilities in permission previews, errors in subagent model resolution, and critical bugs in parallelization and persistence.
Claude Fable 5 has been restored with revised security safeguards and is available until July 7 for paid users, with elevated false positive rates in the initial phase.
Anthropic integrated a hidden tracking mechanism in Claude Code to identify unauthorized resellers and model scrapers — but has already announced its removal.
Autonomous AI code scanners can be tricked by seemingly legitimate instructions in README files to execute malware without triggering classical security checks.
Auto-Mode is now available without explicit opt-in; numerous stability fixes address terminal lags, incorrectly stored remote settings, and faulty session states in agent teams.
Six AI code assistants fail to validate symlinks before write operations, allowing attackers to manipulate system configurations through fraudulent project files — some vendors have already patched, others are still working on fixes.
Six popular AI code assistants (Amazon Q Developer, Claude Code, Augment, Cursor, Google Antigravity, Windsurf) can execute code undetected on developer systems through symlink exploits in attack scenarios known as GhostApproval.
The Claude apps gateway replaces individual credentials per developer with a centralized access point featuring identity, policy, telemetry, routing, and spend-cap functions.
