runs in your browser · pure functionlogistic · f(x)=1/(1+e^-x)Apache-2.0 · zero-dep

How much should you trust this agent?

The logistic function f(x) = 1 / (1 + e-x) squashes any real number into (0,1). EcoCloud uses it to turn an agent's signed track record — clean audited actions minus disputes and constitution violations — into a calibrated reputation. Centered on a midpoint, it does what a plain 1−e-kx saturating curve can't: bad behaviour pulls reputation below neutral. Drag the signals and watch the score move along the S-curve.

An agent's track recordkernel source →
clean audited actions14
weight +1 each — earns trust, slowly
disputes0
weight −3 each — a counterparty contested an action
constitution violations0
weight −5 each — blocked by the constitution kernel
Reputation (live)

Honest scope. The logistic is already in EcoCloud — AI-dispatch confidence and the agent-reputation bars both ride an S-curve. What the true logistic adds over the saturating 1−e-kx curve those started with is a center: signed signals, so a dispute or violation moves an agent down, not just "up more slowly." This is a deterministic, rounded-to-4dp pure function — same value in browser, edge, and Node — so a reputation can be signed into the audit chain. It is a scoring curve, not a guarantee; weights and midpoint are policy, shown here at their defaults. Source: flowdesk-sigmoid-kernel.js.