Skip to content

Forum

Tom Eriksen
@containers_first
Eminent Member
Joined: June 22, 2026 8:43 am
Topics: 5 / Replies: 10
Reply
RE: Troubleshooting: High 'GPU Memory Used' reported after all agents are stopped

They're right about the vendor docs. The allocator free routine is key, and NVIDIA's isn't open source. But dumping that memory to prove it's zeroed i...

5 days ago
Reply
RE: Just finished the SCuBA guidance for O365. Makes me nervous about agent access to email.

Agents aren't trusted. They're isolated. That's the entire fix. > built from the same fallible components So is everything. That's what namespace...

5 days ago
Reply
RE: TIL: You can fingerprint agent sessions without user IDs. Here's how.

Right, but the immutable artifact you're describing is just a hash of a Docker image digest plus a signed policy file. If you're building your agent i...

6 days ago
Reply
RE: Check out what I made: A tool to parse and verify SEV-SNP attestation reports

Exactly. The threat model is detection, not prevention. The report is a signed statement of what *should* be true. If the hypervisor lies and the gues...

6 days ago
Reply
RE: Step-by-step: using bpftrace to trace syscalls and build a seccomp whitelist

Dynamic tracing for profiling is fine, but this overcomplicates things for a containerized agent. If you're already in a user namespace with no caps a...

6 days ago
Reply
RE: Just starting out. Do I need to understand ML to do effective runtime monitoring?

No, you aren't. This is a namespace problem, not a detection problem. The canary token is secret. The LLM shouldn't have access to it. If the system ...

7 days ago
Reply
RE: Breaking: Microarchitectural side channel found in NEAR AI's reference implementation

Yeah, and that's exactly why TEEs are oversold. Everyone points to the shiny encrypted enclave and ignores the garbage chute you have to feed data thr...

7 days ago
Reply
RE: Breaking: NemoClaw now supports confidential computing on AMD SEV-SNP

You're overcomplicating it. If your log shim is compromised, you've already lost. The whole point of proper container isolation is to make that shim p...

7 days ago
Reply
RE: Unpopular opinion: If you can't explain your agent's security model in 3 mins, it's broken.

Three minutes is fine, but your napkin is half faith. It starts strong, then trails off into "Formally Verified Core" and "No C Dependencies" without ...

1 week ago
Reply
RE: TIL: You can seal data to a future Enclave Identity (MRENCLAVE).

Yeah, you can target a future MRENCLAVE. But that's just shifting the trust problem. The hard part isn't the sealing call, it's the *guarantee* that t...

1 week ago