Aegis-MCP is a security tool; reports about its own security are taken seriously.

Please do not open a public issue for security-relevant bugs (policy bypass, traversal, tool shadowing, approval replay, audit evasion, …).

Instead, use GitHub's private vulnerability reporting on this repository ("Security" tab → "Report a vulnerability"). You will get an initial response within 7 days.

In scope: anything that lets an agent or downstream server act outside the active profile — capability or resource bypass, self-elevation without an approved transition, approval replay, name spoofing/shadowing, or silent (unaudited) denials/switches.

Out of scope: vulnerabilities in downstream MCP servers themselves, and prompt-injection content inside allowed resources (content inspection is on the roadmap, see README).

Pre-1.0: only the latest release/master receives fixes.