[ctx]ctxseccontext security: attacks on the harness and context layer of AI agents, plus exploited-now news, hunting and playbooks.

← Threat Hunting · markdown

hunting guide (based on Amazon Threat Intelligence, Nov 20, 2026, and MONAI Label GHSA-qc22-rmgc-vg94)

Hunt: Cairn-driven intrusions: MONAI Label datastore writes outside the datastore, a modified sitecustomize.py, and traffic from 142.4.98.145 / 142.4.107.34

  • Network: search firewall, proxy and flow logs for 142[.]4.98.145 and 142[.]4.107.34 (records dated Sept 8-10, 2026).
  • Web logs: on MONAI Label servers, look for PUT /datastore/image or /datastore/label requests whose image parameter contains ../ or absolute paths.
  • Files: check every Python environment on those hosts for new or changed sitecustomize.py/usercustomize.py and unexpected .pth files.
  • Process: Python processes started by the MONAI Label service that open outbound sockets or spawn shells.
  • Agent fingerprints: long, tireless sequences of Redis replication/module abuse, SSRF probes and credential sprays from one source are typical of autonomous tooling.

Sources: AWS Security Blog (CJ Moses): Chinese-nexus actor's Cairn AI agents found a zero-day on their own (Nov 20, 2026) · MONAI Label GHSA-qc22-rmgc-vg94: Path traversal in datastore upload endpoints allows arbitrary file write (critical, Nov 18, 2026)