I built Weblyr (free, no credit card needed). Highlight text on any webpage. Leave threaded comments. Perfect for design reviews, QA feedback, client work, anything. Works on any site: Figma, staging servers, docs, designs. Try it: weblyr.com

The Node.js Brotli bug now has a working RCE demo on a stock build. Imperva got code execution on the official Node 24.22.0 image. Two ways in: fetch() reading a Brotli response, or Express 5 / body-parser 2 decoding Brotli request bodies. A plain Node HTTP server doesn't decode bodies on its own. It took several tries, and no exploitation reported yet. Fixed in 22.23.4, 24.22.1, 26.12.1. Check your container images, they ship their own node. imperva.com/blog/node2shell-…

musl's regcomp can be broken by the size of a regex, before it matches a single character. Integer overflows, present since 0.5.0, all versions through 1.2.6. Crash at least, maybe heap writes. Only hits you if untrusted users can supply patterns. Think search boxes, filters, rule engines on Alpine images. No 1.2.7 yet, only a patch. openwall.com/lists/musl/2026…

Laravel shipped the fix for a 9.2 auth bypass on Aug 25. The release notes list one change: "Fix resource loading." The advisory came out Oct 19. Eight weeks later. No CVE. If a guard in config/auth.php uses driver 'token' on MySQL or MariaDB, an attacker can log in as an existing user without knowing their token. Fixed in 12.68.0 and 13.29.0. Rotating tokens won't help. Update. github.com/laravel/framework…

Lazarus is now hiding its payloads with post-quantum crypto (ML-KEM-1024). Kaspersky says they adopted it faster than many security products. But the crypto isn't how they got in. A phishing link to GitHub Pages, then a mandatory Korean banking security app, with the GitHub API as the C2. The tell is a built-in Windows binary (fsquirt.exe) running under a security service. Encryption can't hide where code runs from. securelist.com/lazarus-opera…

The Browser Company is shutting down Arc and says the reason is security. Keeping a browser safe is getting harder with AI, they say. Everything moves to Dia. New accounts stopped Oct 22. Desktop stops working Dec 22, and account data is deleted within 30 days after. Export your passwords and Spaces now. arc.net/

7 of 10 HTTP/3 stacks FoxIO tested had the same bug: tiny header block, huge memory. quic-go fixed it in 0.57.0. It sits under Caddy, Traefik and cloudflared, so check your build, not your config. Bun's HTTP/3 is still unpatched. One 516 KiB stream pins it at 100% CPU. foxio.io/blog/header-accordi…

Arista VeloCloud Orchestrator has a zero-day that's already being exploited. Any logged-in user, even the lowest role (Enterprise Read Only), can become a superuser operator. So your "harmless" read-only accounts are admin keys now. The web UI is exposed by default and no config turns that off. Save the logs first, then patch: 5.2.3.18, 6.1.3.10 or 6.4.2.12. Block 74.82.197.86 and 38.54.50.151. arista.com/en/support/adviso…

Microsoft says the ConnectWise cloud attack was run with an AI pentest tool called ARTEX. In several cases it took under 3 hours from first target to privileged access. Most attempts failed, but your "we'll look at it Monday" window is gone. If you ever authorized ConnectWise apps in Entra, search by app ID, not display name. Disable, don't delete, until the logs are saved. microsoft.com/en-us/security…

Quad9 is leaving France on Nov 24 instead of blocking domains for a court. It says it can't tell which users are in France, so it can't block just them, and it can't keep paying the fines. About 20 more suits want the same thing. If your French users or servers lean on 9.9.9.9, check your fallback resolver now. quad9.net/news/blog/au-revoi…

Your MCP server picks the OAuth login URL. On Windows, OpenCode passed it to PowerShell without escaping. So a server you configured can run commands as you during sign-in. Windows/WSL only, fixed in 1.18.33. No CVE, so scanners won't flag old versions. Vendor says a bad MCP server is out of scope. pillar.security/blog/open-an…

eMudhra is quitting public TLS certificates. Their release blames low revenue and says the infrastructure wasn't compromised. Technically true, nobody stole anything. Eight of their intermediate CA keys were just one byte long, from a bug in their key ceremony. Their G1 roots keep running during the transition, so keep your CA bundles updated. emudhra.com/en/press-release…

Trend Micro's ZDI mailed a ksmbd bug report to [email protected] and CC'd the public kernel list. Full PoC attached. Kernel SMB server, no login needed, CVSS 9.0. The report says it needs a guest share and a race. If you run ksmbd, check your shares for guest access today. lore.kernel.org/lkml/BYAPR01…

LiteLLM's attacker didn't skip the approval step. They had the approver. Per LiteLLM, one engineer's account edited the release workflow and started the run. A second account approved the production gate. A third deleted the run records. It looks like stolen GitHub CLI tokens, and they still don't know how they were taken. Two-person approval means little when one attacker holds both people's tokens. docs.litellm.ai/blog/securit…

Railway is the second app platform this month to lose customer env vars. Zeabur was Oct 14. Railway says an attacker may have read encrypted variables, and for some, the decrypted values. It thinks the end goal is AI inference providers, so LLM API keys are the loot. Rotate at the provider, not just in the Railway dashboard. Then check usage on those keys. blog.railway.com/p/security-…

Someone tried to exploit PHP2Shell on Oct 17. The patch shipped Oct 21. Amazon saw it on a honeypot, 4 days before disclosure. They don't know how the attacker got the details. So if you only check logs from the patch date, you're looking at the wrong window. Start at Oct 17 and search for POSTs with PHP_SESSION_UPLOAD_PROGRESS. aws.amazon.com/blogs/securit…

An OpenAI agent deleted its own safety monitor during a code migration. Asked if it did, it said no. A second agent was told "don't send Slack messages" and used that to switch off the alerts meant for humans. OpenAI says it has no satisfying explanation yet. If an agent can edit its own monitoring, its "task done" message is not a log. Keep the log somewhere it can't reach. alignment.openai.com/misalig…

Two critical PHP bugs were kept quiet for two months while OpenAI, Wiz, Cloudflare and government agencies prepared fixes and WAF rules. One of them was found twice, by Calif and by Wiz, working separately. That's the problem with embargoes now. If AI finds a bug for one team, it can find it for another. openai.com/index/coordinated…

Your self-hosted GitLab may be one regex away from a full takeover. CVE-2026-110264, CVSS 10. Anyone on the network can send a crafted regex that overflows an integer in the compiler and runs code on the server. Every version since 7.0 is affected. GitLab says it needs "certain conditions" and lists no workaround. That box holds your source and your CI secrets. Patch to 19.5.1, or your branch's fix (18.11.13 for anything from 7.0 to 18.11). docs.gitlab.com/releases/pat…

The eMudhra CA keys weren't stolen. They were one byte long. A P-384 private key should be 384 random bits. These were 47 zero bytes and one number under 256. Scott Helme cracked them in under a second on a laptop. eMudhra says a ceremony script bug made a second, weak key pair on the HSM, and someone picked the wrong one. Live for 17 months. scotthelme.co.uk/eight-bits-…

LiteLLM 1.107.0 on PyPI is backdoored, and you don't even have to import it to get hit. It drops cook.pth into site-packages. Python runs .pth imports at startup, so every Python process on that machine runs the stealer. It grabs SSH keys, cloud creds and env vars, then leaves a daemon that pulls new code every hour. Installed it? Rotate everything that host could see. safedep.io/litellm-pypi-comp…