UTC

AML.CS0070

Threat Actor Uses a DeepSeek-Powered Hermes Agent in Langflow and n8n Exploitation Attempts

Type
Incident
When
2026-05-07
Actor, as ATLAS names it
Not stated
Target
Publicly exposed Langflow and n8n systems, primarily in China
  • ATLAS attributes it to knaithe (adversarial operation).
  • Reported by Unit 42.
  • On ShieldMap: Langflow (matched by name in the target).
Techniques, in order
  1. 1Obtain Capabilities: Generative AIResource Development · AML.T0016.002
  2. 2Obtain Capabilities: Software ToolsResource Development · AML.T0016.001
  3. 3Obtain Capabilities: AI Agent ToolsResource Development · AML.T0016.004
  4. 4Develop Capabilities: AI Agent ToolsResource Development · AML.T0017.002
  5. 5AI Model Inference API AccessAI Model Access · AML.T0040
  6. 6Autonomous Attack-Path AdaptationAI Attack Adaptation · AML.T0117
  7. 7Generate Malicious CommandsAI Attack Adaptation · AML.T0102
  8. 8Autonomous ReconnaissanceReconnaissance · AML.T0116
  9. 9Search Open Technical Databases: Scan DatabasesReconnaissance · AML.T0000.003
  10. 10Obtain Capabilities: ExploitsResource Development · AML.T0016.003
  11. 11Active ScanningReconnaissance · AML.T0006
  12. 12Exploit Public-Facing ApplicationInitial Access · AML.T0049
  13. 13Search Open Websites/Domains: Code RepositoriesReconnaissance · AML.T0095.000
  14. 14Search Open Technical DatabasesReconnaissance · AML.T0000
The full case study on MITRE ATLAS →

From MITRE ATLAS 2026.10 (Apache-2.0). ShieldMap lists who and what was involved and the techniques used, and links to ATLAS for the write-up; it does not copy the steps. Attribution is ATLAS's, not ShieldPi's.