Cursor IDE Auto-Executes Malicious Code in Poisoned Repos
2026-07-14T14:51:41Z•6d666bbaab97c461689d3d40cb2a9ceb6978d00dae2d7365443b02c6163c929f
active-exploitationagentic-workflowsagentjackingai-securitycloud-securitycode-executioncritical-infrastructureidentity-and-access-managementincident-responseinfostealerkernel-driver-abusellm-ransomwaremalvertisingphishingpoisoned-reposransomwaresigned-kernel-driverssupply-chainvulnerabilitieszero-day
What happened
Major themes across these items: AI development and agent tooling are emerging attack vectors — multiple flaws and techniques (Cursor auto-execution from poisoned repos, agentjacking, GitLost, Dialogflow CX rogue agents, LLM-driven ransomware like JadePuffer, and AI-gateway exposures) enable supply-chain and code-execution attacks against development workflows and automated agents. Active exploitation and proof-of-concept releases continue for high-impact products (Windows Defender 'RoguePlanet' PoC, Citrix NetScaler memory disclosure, FortiBleed compromises and Nextcloud zero-day), while new/
Why it matters
A reviewed impact interpretation has not been published for this record.
Evidence and limitations
- Source ID
- darkreading
- Record identifier
- 6d666bbaab97c461689d3d40cb2a9ceb6978d00dae2d7365443b02c6163c929f
- Enrichment time
- 2026-07-14T14:51:41Z
- AI-assisted enrichment
- Yes
This record may overlap with other records. Its enrichment can be incomplete or wrong, and machine assistance was used. Validate consequential decisions against the linked source and your own environment.