🧰 Mythos Preview

The items that name it 2

• Have We Seen an Acceleration in Discoveries?
  article2026-08-14

  Anthropic model run with Claude Code to find vulnerabilities and write working exploits from a generic prompt

• IronCurtain: Secure AI Agent Runtime
  writeup2026-05-07

  Anthropic's restricted frontier model reported to discover novel vulnerabilities, including the 1998 OpenBSD TCP SACK flaw

In every workspace you can read 4