Compare AI Agent Blueprints

Choosing between two agent designs? Benchmarks won't tell you which is safer to run. This puts two blueprints side by side on what actually decides it: their AgentAz Trust Level, how many tools auto-execute versus wait for human approval, the declared worst-case action, and the practical setup. Every value is read from each blueprint's spec — nothing is scored.

How to read the comparison

The most important row is Trust Level: it summarizes how much authority an agent holds without a human (ADV is advisory; A5 is fully autonomous). Two blueprints can do the same job at very different autonomy — one gating its consequential tools, another letting them auto-execute. The auto-executing vs. approval-gated tool counts show exactly where that line falls. Lower autonomy and more gated tools aren't “better” in the abstract — they're safer, and the right trade depends on how reversible the agent's actions are. To go deeper on either one, open its page for the full spec, or scan your own agent to see where it lands.