Threat hunts for PyPI

8 hunts covering PyPI, each with a hypothesis, the queries that test it and what the hunt cannot see.

8 hunts, newest first

All 486 hunts
  1. Agent-Framework Exfiltration to Webhook.site

    An intruder is exfiltrating harvested cloud and developer credentials from Python runtimes to a malicious webhook.site endpoint following the agent-framework PyPI compromise.

    High Elastic Security Labs Part 2 of 2 · 3 queries, 8 steps credential access, execution
  2. Malicious Python Package Persistence and Secret Theft

    An intruder has installed a compromised version of the agent-framework package which persists via a .pth hook and harvests cloud credentials by reading environment variables from the /proc filesystem and local configuration files.

    High Elastic Security Labs Part 1 of 2 · 4 queries, 11 steps credential access, execution
  3. LiteLLM Supply Chain Poisoning and Persistence

    An adversary has installed a poisoned version of the LiteLLM library to establish persistence via Python startup files and exfiltrate host credentials.

    High ESET Research Part 2 of 2 · 4 queries, 11 steps CVE-2026-109618, CVE-2026-109619
  4. LiteLLM Remote Code Loading and Exfiltration

    An adversary is using malicious Python startup hooks in a compromised package to execute a remote code loader and exfiltrate credentials from developer and production environments.

    High Microsoft Part 2 of 2 · 3 queries, 8 steps credential access, execution
  5. LiteLLM 1.107.0 Supply Chain Compromise and Credential Harvesting

    An intruder has distributed a compromised version of LiteLLM (1.107.0) that uses Python startup hooks to execute a credential harvester upon any Python invocation, targeting cloud, SSH, and browser secrets.

    High Microsoft Part 1 of 2 · 3 queries, 8 steps credential access, execution
  6. ChocoPoC: Mapbox Dead-Drop C2 and Exfiltration

    An adversary is using trojanised Python packages to establish C2 via DoH and Mapbox datasets on researcher workstations, subsequently exfiltrating credentials from local password stores.

    High Sekoia Part 2 of 2 · 3 queries, 8 steps CVE-2025-55182, CVE-2025-64446
  7. Linux Fileless and In-Memory Execution

    An adversary is executing malicious code on Linux hosts by staging payloads in memory-backed file descriptors, using interpreter one-liners, or running unlinked binaries to avoid on-disk detection.

    High Elastic Security Labs 5 queries, 11 steps defense evasion, execution
  8. IT Support Impersonation and WinRM Lateral Expansion

    An adversary has hijacked a remote-support session to execute PowerShell, use a portable Node.js runtime for C2, and expand laterally via WinRM to domain controllers.

    High Microsoft 5 queries, 11 steps command and control, discovery

Hunt faster than the attacker.

Huntbase is the autonomous threat-hunting platform. When new research lands, Scout checks back through your history across the tools you already run, responds within the rules you set and verifies the threat is gone.

Content pack

Sign up to open

Pack contents open in Huntbase.