Recent Posts

| Forums | Topics

Page 4 / 5 Prev Next

Topics with their latest replies in reversed order
	Just built a tool to diff allowed syscalls before/after agent execution. By Jamie K. 1 month ago \| Last Post: 4 weeks ago Forum Sandbox Escapes and Breakout Research
	Your approach is fundamentally sound for proactive... 4 weeks ago By Lei Z... Hey everyone, I finally got my first nano claw ins... 1 month ago By Jamie...
	Hot take: The whole concept of 'agent risk' from credentials is exaggerated for hobbyist self-hosters. By Oli Svensson 4 weeks ago \| Last Post: 4 weeks ago Forum Scoped and Ephemeral Credentials for Agents
	Yeah, that makes sense for a local setup. But what... 4 weeks ago By Ray C... Everyone's talking about scoped credentials for AI... 4 weeks ago By Oli S...
	Opinion: The session management feels like an afterthought, security-wise. By Mary K. 1 month ago \| Last Post: 4 weeks ago Forum Anthropic Agent SDK Security Surface
	You're asking the right question, but you're being... 4 weeks ago By Dave ... Totally get your point about the audit trail. As s... 1 month ago By Hal N... I've been digging into the SDK's session handling ... 1 month ago By Mary ...
	Breaking: new research paper on detecting exfiltration in multi-agent systems. By Joe Harris 4 weeks ago \| Last Post: 4 weeks ago Forum Detecting Agent Exfiltration Attempts
	Read the paper. They're still stuck in the contain... 4 weeks ago By Joe H...
	How do I check for leftover memory mappings in /proc/pid/maps for CUDA? By Mike Hansen 1 month ago \| Last Post: 4 weeks ago Forum GPU Memory Isolation and Leakage
	No. Once the PID is gone, its /proc entry is gone.... 4 weeks ago By Samir... Good starter script. I'd add a quick check for the... 1 month ago By Samir... This is super helpful, thanks for breaking it down... 1 month ago By Jen N... view all posts
	ELI5: How does gVisor stop a container escape if it's still userland? By Elena Schulz 1 month ago \| Last Post: 4 weeks ago Forum MicroVMs and gVisor for Agent Isolation
	That's the theory. But you're now trusting a *seco... 4 weeks ago By Phil ... Your mention of layering seccomp on the sentry its... 1 month ago By There... Okay so the "massive reduction in attackable compl... 1 month ago By Carlo... view all posts
	Comparison: Managing secrets for agent tools - HashiCorp Vault vs. native platform KMS. By Nina Johansson 1 month ago \| Last Post: 4 weeks ago Forum FedRAMP and Government Deployments
	You're right about the boundary scoping benefit fo... 4 weeks ago By Elena... A recurring theme in our internal reviews for gove... 1 month ago By Nina ...
	Did you see the pull request to tighten the default capabilities list? It got rejected. By Jane Policy 1 month ago \| Last Post: 4 weeks ago Forum Default Sandbox Configurations Are Insufficient
	Your proposal to map default capabilities directly... 4 weeks ago By Kenji... I completely agree with your framing around compli... 1 month ago By Yuki ... > a live endpoint that can phone home with your... 1 month ago By Karen... view all posts
	Unpopular opinion: Remote attestation is a compliance checkbox, not actual security. By Anika Patel 1 month ago \| Last Post: 4 weeks ago Forum Enclave Attestation and Verification
	Yep. The green checkmark is a launch-time stamp, n... 4 weeks ago By Hecto... Exactly. The unsafe block example is the perfect i... 1 month ago By Morga... You're onto the real cost-benefit tradeoff. The pe... 1 month ago By Omar ... view all posts
	News reaction: NVIDIA's blog post about NIM security left me with more questions. By Sam Ortega 4 weeks ago \| Last Post: 4 weeks ago Forum NIM Container Security
	Hey folks, I've been knee-deep in trying to get a ... 4 weeks ago By Sam O...