An AGENTS.md in your Downloads folder: how a file can take over a coding agent
Researchers found a malicious AGENTS.md in a real Downloads folder that rewrote Codex's own config. How the attack works and what to check on your machines.
5 min read
AI agents that plan and act: coding agents, background agents and multi-agent setups.
Researchers found a malicious AGENTS.md in a real Downloads folder that rewrote Codex's own config. How the attack works and what to check on your machines.
5 min read
The filesystem, everything and fetch reference servers fix eight advisories. Why date-versioned installs must upgrade by hand, and what Windows still lacks.
A VS Code flaw let a Copilot-harness agent add a shell command to a hook file without asking. What was fixed, how the ratings differ, and what to check.
4 min read
CERT-In says agents under evaluation reached Indian firms' live systems through exposed credentials. What the advisory asks of teams that run agents.
5 min read
New research finds low attack success rates mostly mean agents ignored injections, and detector rankings don't transfer. How to read these evaluations.
5 min read
FastMCP published 16 advisories in five days, including client-side code execution. The versions that fix them, and the steps an upgrade won't do.
A large MCP server now lets agents connect with minimal OAuth scopes and ask for more mid-task. How the specification's step-up flow works, and what to check.
5 min read
GitHub's package firewall checks npm installs from CI runners and coding agents against age, malware and attestation rules. How it works, and its limits.
6 min read
Anthropic built Scion after daily merges rose from 2,500 to over 30,000. What drove it, what it changed, and which lessons apply to teams that stay on Git.
Beam arrived on 16 October under Apache 2.0 with a long safety report. Its agent results show which safeguards come with the weights and which don't.
Manifold checked every URL in public agent skills, MCP servers and plugins; 121 components use a high-risk domain. What that means for reviewing skills.
ATT&CK v20 gives direct and indirect prompt injection, AI-enabled exfiltration and two related behaviours their own IDs. What changed, and the gaps to plan for.
6 min read