Your agent reads untrusted content and can send data externally. How do you stop prompt-injection data exfiltration?

When an agent holds private data, reads untrusted text, and can communicate out, injected instructions can steal data. This 'lethal trifecta' is the defining agent vulnerability. The fix is architectural, not a better prompt.

Updated Aug 2026 · Grounded in real Applied AI Engineer interview loops and written to a senior-engineer editorial bar.