Resources
AI Agent Authorization Resources
Technical guides on pre-execution control, OWASP agentic security, and Verification Before Execution™ for AI payment and treasury workflows.
AI Agent Authorization vs. Observability: Why Watching Isn't Stopping
Observability tells you what happened. Authorization decides whether it should happen at all. Why observability alone is structurally insufficient for autonomous agents taking high-value actions.
Why Audit Logs Are Not Pre-Execution Controls
An audit log is a historical record, not a pre-execution gate. For high-consequence AI agent actions, a log entry is a receipt, not a safeguard. What RF-100 requires at the control point.
How to Stop AI Agent Self-Approval: A Technical Guide
Self-approval occurs when an AI agent uses its own context to authorize the high-value actions it is also requesting. How independent verification authority eliminates it.
MCP Tool Authorization and Execution Control
MCP has no built-in mechanism for pre-execution proof that a tool invocation was independently authorized. How ExecutionProof wraps MCP tool calls with ALLOW/HOLD/DENY control.
OWASP Agentic Security Controls — Pre-Execution Implementation
Mapping the OWASP Top 10 for Agentic AI to RF-100 requirements and ExecutionProof enforcement. OWASP identifies what can go wrong; RF-100 specifies what must be verified before execution.