Attackers can inject indirect prompts in normal-looking repositories to trick Claude Code into spawning a reverse shell.
GLM-5.2, Z.ai’s open-weight model, has reached 39% F1 on Semgrep’s IDOR benchmark, beating Anthropic’s Claude Code coding assistant in the prompt-only lane. Claude Code scored 37% F1 with Opus 4.6 and ...
Ongoing research into AI agent framework security identified an exploit chain in AutoGen Studio (AutoGen’s open-source prototyping user interface) that allows untrusted web content rendered by a ...
DeepReinforce today released Ornith-1.0, a family of open-source coding models built around a mechanism most RL-trained agents avoid: the model itself writes the training harness that guides its own ...
GitLab CE/EE security updates resolve 13 vulnerabilities, including high-severity code execution and information disclosure ...
Shield detects your tech stack, runs every applicable security scanner in parallel, consolidates findings into a single report, calculates a risk score, proposes code fixes, and optionally files ...
AI dev tooling is the Wild West, and APM is the only credible attempt at a cross-agent dependency manager. Composable flavour packages mean you pull in only what each project needs. Meanwhile, the ...