The research collection

Security in working systems.

Documented failures, constructive defenses and careful readings of the evidence. Start with the access decision behind each case.

Your agent can have permission to do the wrong thing

Which boundary separates valid access from authorized disclosure? A reported GitHub MCP case and four layers of authority.

The recurring question

What can go wrong when I give this agent access—and what should I change before trusting it?

Read access, write access and remote authority are different boundaries. A constructive review of what a sandbox can—and cannot—establish.