Learn to secure
agent actions
In-depth, evergreen writing for the people wiring agents into real systems - and the people accountable when they act. For timelier notes, see the Blog.
Coming next
Guide· Coming soon
A threat model for tool-using agents
What actually goes wrong when an agent has credentials - from prompt injection to tool abuse to data exfiltration.
Guide· Coming soon
Designing human approval that people don't hate
Approval flows that catch what matters without turning your on-call into a click farm.
Guide· Coming soon
Short-lived credentials for autonomous systems
Why standing service keys and agents don't mix - and what scoped access looks like in practice.