Mini Shai-Hulud is a campaign conducted by TeamPCP. TTPs were influenced by the original Shai-Hulud campaigns, which have not been attributed to TeamPCP.
Don’t Overanalyze AI-Generated Payloads
- RU/CIS check is likely an artifact — “AI please generate me malware, make no mistakes” produces boilerplate geo-checks, not a useful attribution signal
- Wipers are more stunt than substance — the earlier Kamikaze.py and the current “roulette” wiper targeting specific geos can do real damage, but shouldn’t be your headline
Report Card: Code Quality
Multiple bugs in the payloads have substantially reduced impact:
- MistralAI & UiPath npm:
setup.mjsreferencestanstack_runner.js, but the actual payload is namedrunner_init.js. Mistral’s advisory confirmed the npm packages “contained inoffensive code with a non-functional reference file” - transformers.pyz v1: In the first version of the PyPI payload cluster, the wiper was broken
Exposure Windows Are Shrinking
npm, PyPI, and maintainers have clearly invested in faster response to community reports:
| Target | Exposure Window | Source |
|---|---|---|
| TanStack | ~6 min to publish, detected in ~20 min (2026-05-11 19:20–19:26 UTC) | Postmortem |
| Mistral AI (npm) | ~3 hours (2026-05-11 22:45 – 2026-05-12 01:53 UTC) | Security Advisory |
| UiPath | 65 packages, removed same day | Trust Center |
Victim Response
TanStack, Mistral AI, and UiPath have all shown prompt and responsible incident handling. Each has published a full postmortem publicly:
- TanStack Postmortem — detailed attack chain analysis, follow-up hardening
- Mistral AI Security Advisory — MAI-2026-002, affected versions, remediation steps
- UiPath Trust Center — ongoing investigation, cloud remediation deployed