Self-hosted, Self-Improving, Federated Everything App — dev management, agent orchestration, persistent mind, digital twin, second brain, and creative pipeline from a single dashboard over Tailscale to port the user into the OS
JavaScript ai-agentsdashboarddeveloper-toolsdigital-twinquantified-selfsecond-brainself-hostedtailscale
/**
* How a measured local-model assessment describes a launch configuration that
* did NOT take effect.
*
* `tuningApplied === false` covers two different failures, and they read as
* opposite things to the person looking at the row:
*
* - a TUNED run whose knobs never reached the daemon — the numbers describe
* some other configuration, not the one in the label;
* - an UNTUNED run PortOS could not put back on backend defaults, because the
* daemon was still carrying the previous run's tuning. Its label already
* says "backend defaults", so "tuning was not applied" reads as a
* contradiction unless the copy names what actually happened.
*
* `tuningKey` is what separates them: `''` is the untuned run.
*
* This is the SHORT form, for a dense table cell or a toast suffix. The full
* sentence for a surface with room to say it is the server-side exclusion reason
* in `getAssessmentReport` — keep the two in step.
*
* @returns {string|null} `null` when nothing is wrong with the reading.
*/
export function tuningNoticeChip(entry) {
if (entry?.tuningApplied !== false) return null;
return entry?.tuningKey ? 'tuning not applied' : 'not at defaults';
}