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
import { useState, useEffect, useMemo } from 'react';
import { getSettings } from '../services/api';
import { resolveRenderCfg } from '../lib/pipelineImageDefaults';
import { deriveAvailableBackends } from '../lib/imageGenBackends';
/**
* Load the settings blob once on mount and expose "what will this record's next
* render actually run on" as a ready-to-use `imageCfg`. Collapses the
* `getSettings` + `resolveRenderCfg` pair every single-image-render call site
* re-implements (Story Builder's characters step, the universe base-style
* probe, the Decks render bar). The resolution itself lives in
* `resolveRenderCfg` — read its header for the ladder and for why nothing here
* may read `settings.pipeline.imageGen`.
*
* Components that already load the full settings blob for other reasons (e.g.
* the Universe Builder reads loras + models from the same fetch) should call
* `resolveRenderCfg` against that shared fetch rather than double-fetching here.
*
* @param {object} [opts]
* @param {object|null} [opts.record] - Record whose `imageMode`/`imageModelId` pin wins.
* @param {string|null} [opts.target] - RENDER_TARGET id whose `renderDefaults` pin is next.
* @param {boolean} [opts.includeExternal] - Also list an enabled external SDAPI
* backend. Default false; the mood-board poster opts in so the user can pick
* any configured image service.
* @returns {{ imageCfg: object, backends: Array<{id:string,label:string}> }} —
* `backends` is the enabled list the pin ladder was gated on (external omitted
* unless `includeExternal`; empty until settings land), for callers that also
* render a backend picker.
*/
export default function useImageRenderSettings({ record = null, target = null, includeExternal = false } = {}) {
// `null` = not fetched yet (or the fetch failed), which is NOT the same as a
// settings blob with no backends enabled. Holding the raw blob (rather than a
// derived cfg seeded with the defaults and an `[]` backend list) keeps the two
// apart: `[]` reads as "loaded, nothing enabled" and would suppress every pin.
const [settings, setSettings] = useState(null);
useEffect(() => {
getSettings({ silent: true }).then(setSettings).catch(() => {});
}, []);
// Depend on the pin fields rather than the record identity — callers hand us
// a freshly-fetched draft object on every save, and re-deriving an identical
// cfg would churn the render opts for every consumer downstream.
return useMemo(() => {
const backends = settings ? deriveAvailableBackends(settings, { excludeExternal: !includeExternal }) : [];
return {
backends,
// Display projection only; tagged submissions retain server inheritance.
imageCfg: resolveRenderCfg(settings, { record, target }),
};
}, [settings, target, includeExternal, record?.imageMode, record?.imageModelId]);
}