the scariest part of the carmaker ForgeRock research by @samwcyo and @infosec_au isn't the RCE. it's that none of it needed a ForgeRock 0-day. it came down to hardening the vendor had already documented.
codex-security by @OpenAIDevs shipped v0.10.0 with a --max-total-cost cap for batch runs. useful, since agent scans on a big repo get expensive fast.
the mode worth putting in CI is diff scanning (cs scan . --diff origin/main) with SARIF output.
https://t.co/1cz1Ld5ci3
@Jhaddix@Xbow unauth RCE on an election commission app right after midterms is wild. curious if it was a known CVE or something the agents found from scratch
@hakluke small single-purpose tools make so much sense for agents. curious how you handle scope though, hard block out-of-scope hosts in the tool wrapper or just trust the prompt?
@TheHackersNews the chrome extension route is the sneaky one. a malware download at least has a shot at tripping EDR, a search hijacker just sits in the browser for months
@TheHackersNews the console bypass worries me more than the 9.5 DoS tbh. the console manages your whole ADC fleet, so one auth bypass there is a lot of blast radius
patched but still exploitable is the worst kind of advisory. if your N-central appliance faces the internet, put it behind a vpn/allowlist today and don't wait for the fixed fix. one RMM box = every customer network behind it.
‼️ BREAKING: Recently patched N-central servers remain exploitable under certain conditions.
"4EverAble2Shell" enables unauthenticated root RCE; "Agent2Shell" can run Windows SYSTEM commands with the appliance ID.
Attackers are probing. N-able is investigating.
Details: https://t.co/t3tenC9MPi
telegram opened HTML widgets to all bots on Nov 9. two days later researchers showed code execution on Android from a bot message via the widget JS engine.
a new parser in every chat is new attack surface. it's patched and widgets now run in an isolated process. update: 13.2.2.
@watchtowrcyber default config again 😬 netscaler memory bugs have a habit of going from advisory to mass exploitation in days, patch tonight not next sprint
@TheHackersNews "patch blocks the exploit but not session hijacking" is the line that matters here. MSPs patching and calling it done are gonna have a bad time 😅
@TheHackersNews approving the real MFA prompt doesn't help when the proxy just walks off with the session cookie. push MFA really isn't built for AiTM, passkeys are
@vxunderground heaven's gate in 2026 is a flex lol. curious if it drops the final payload entirely when it sees mssense or just switches the delivery path
@TheHackersNews worth noting the PoC basically forges a call to an existing Server Action with your args, so how bad it gets depends on what actions the app ships. one more reason to treat actions as public endpoints