Skip to content

Threat Intelligence Feed

LIVE

Live indicators of compromise for AI agent security. Updated continuously from OpenSyber Research, NVD, and CIRCL.

Total IOCs

4

Last Updated

0m ago

Feed Sources

4

Auto-Block Rules

2

critical

Malicious dependency campaign targeting AI coding agents

AUTO-BLOCK

OpenSyber Research is tracking package impersonation and credential harvesting against agent workspaces in the United States and Germany.

supply-chaincredentialUnited StatesGermany
opensyber-research0m agocampaign
high

Prompt injection attempts against repository automation

AUTO-BLOCK

Observed prompts attempt to bypass MCP policy and exfiltrate secrets during automated pull-request review.

prompt-injectionblockedIndia
community0m agotechnique
medium

Container escape advisory for untrusted tool runners

MONITOR

A runtime hardening advisory for teams running shell and browser automation inside contractor workspaces.

containerBrazil
nvd0m agoadvisory