Security Advisories
Vulnerability research and advisories.
Findings from the TraceAgentrix research team on AI apps, agents and MCP servers.
| ID | Severity | Title | Published | Status |
|---|---|---|---|---|
| TA-2026-014 | Critical | Prompt injection leading to tool-call hijack in popular MCP server | Oct 30, 2026 | Patched |
| TA-2026-013 | High | Credential exposure in AI coding assistant config files | Oct 21, 2026 | Mitigated |
| TA-2026-012 | High | Unauthenticated MCP endpoint exposes local filesystem | Oct 08, 2026 | Patched |
| TA-2026-011 | Medium | Clipboard data forwarded to third-party AI plugin | Sep 24, 2026 | Monitoring |
| TA-2026-010 | Medium | Over-permissive OAuth scopes in AI browser extension | Sep 09, 2026 | Patched |
| TA-2026-009 | Low | Verbose telemetry in desktop AI client leaks file paths | Aug 19, 2026 | Patched |
Sample content for design preview only.
Ready to see every AI data flow?
Book a 30-minute walkthrough with our security engineers.