Articles by lebovic
126

Anthropic requires 30 day data retention for Fable and Mythos (claude.com)

3

White House Memo on Adversarial Distillation of American AI Models [pdf] (whitehouse.gov)

2

Benchmarking open-weight models for security research (dualuse.dev)

1

How do frontier AI agents perform in multi-step cyber-attack scenarios? (aisi.gov.uk)

4

Hacking Claude Code remote: escaping YOLO-mode sandboxing (noahlebovic.com)

10

Evaluating and mitigating the growing risk of LLM-discovered 0-days (anthropic.com)

1

Show HN: An agent sandboxing quickstart based on Claude Code (github.com/lebovic)

3

Delay-Line Memory (wikipedia.org)

185

Claude Advanced Tool Use (anthropic.com)

1

Claude Skills is context management for code execution (noahlebovic.com)

3

Fixing a Milli-q purifier for 99% off (bsiranosian.com)

4

Infrastructure from mining makes sense for scientific computing (noahlebovic.com)

1

Barnstorming (wikipedia.org)

4

Show HN: Swap between local and cloud execution with a Python decorator (github.com/trytoolchest)

6

AWS has a low elasticity ceiling for big servers (noahlebovic.com)

144

AWS doesn't make sense for scientific computing (noahlebovic.com)

5

Show HN: Lug – run Python locally or in the cloud, paired with a container