- agent-alpha -> agent-beta -> agent-gamma -> agent-alpha
The default authored cluster fixture contains a three-agent cycle and assigns pre-patch entropy point eight nine and post-patch entropy point one two. At three hundred pixels per fixture unit the bars span two hundred sixty seven and thirty six. Staging the patch changes the displayed cycle count to zero and quorum label to met. But the entropy values come from fixed scenario data, not actual messages or a measured convergence algorithm. The original offline graph library fails; native staging can still update its local labels and manifest. Neither label change proves a running agent cluster was corrected. Entering maximum hops three or ten records different manifest values but both staging calls immediately mark the fixture patched, cycles broken, and quorum met. At twenty pixels per entered hop the bars grow from sixty to two hundred while the same success branch runs. This source does not enforce a real message-hop limit or prove that three hops are sufficient. The chosen mitigation action and commit phase are also recorded settings; the convergence label does not depend on successful remote deployment. Inspect the exported local JSON or YAML as a staging description, not execution evidence. The scrub handler rounds the slider value times point four five. Native slider positions zero and one hundred therefore display zero and forty five seconds; at five pixels per displayed second the bars span zero and two hundred twenty five. Scrubbing changes only this time label, not the graph state or a replayed trace. A separate countdown starts at eighteen then repeats forty five. Staging marks convergence immediately rather than waiting for that clock. The short fixture remediation hash is a supplied string, not a freshly computed cryptographic digest of the exported artifact. These independent local labels should not be treated as deployment, audit, or consensus proof.