Security Advisories

Vulnerability research and advisories.

Findings from the TraceAgentrix research team on AI apps, agents and MCP servers.

IDSeverityTitlePublishedStatus
TA-2026-014CriticalPrompt injection leading to tool-call hijack in popular MCP serverOct 30, 2026Patched
TA-2026-013HighCredential exposure in AI coding assistant config filesOct 21, 2026Mitigated
TA-2026-012HighUnauthenticated MCP endpoint exposes local filesystemOct 08, 2026Patched
TA-2026-011MediumClipboard data forwarded to third-party AI pluginSep 24, 2026Monitoring
TA-2026-010MediumOver-permissive OAuth scopes in AI browser extensionSep 09, 2026Patched
TA-2026-009LowVerbose telemetry in desktop AI client leaks file pathsAug 19, 2026Patched

Sample content for design preview only.

Ready to see every AI data flow?

Book a 30-minute walkthrough with our security engineers.