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 { useEffect, useState } from 'react';
import { Palette, Check, Dice5, Cpu, Settings as SettingsIcon } from 'lucide-react';
import { randomSeed } from '../../lib/genUtils';
import BackendChipStrip from '../media/BackendChipStrip';
import AutoSizeTextarea from '../ui/AutoSizeTextarea';
import { IMAGE_GEN_MODE } from '../../lib/imageGenBackends';
import StagePromptModelPicker from './StagePromptModelPicker';
import { STYLE_ID, EMPTY_IMAGE_STYLE } from '../../lib/wrImageDefaults';
const SCRIPT_STAGE = 'writers-room-script';
function groupPresetsByCategory(presets) {
const map = new Map();
for (const p of presets) {
const cat = p.category || 'Other';
if (!map.has(cat)) map.set(cat, []);
map.get(cat).push(p);
}
return Array.from(map.entries());
}
// World style picker — dropdown of curated presets + Custom + None.
// Selecting a preset fills the prompt textarea; the user can edit it freely
// from there, which flips the presetId to 'custom' the moment text diverges.
// Saves are debounced into a single PATCH on blur (not per-keystroke) — the
// onChange contract is "give me the next imageStyle" not "save it now."
function WorldStyleRow({ value, presets, onChange }) {
const [draftPrompt, setDraftPrompt] = useState(value.prompt || '');
const [draftNeg, setDraftNeg] = useState(value.negativePrompt || '');
// Pull the saved value down when the work id swaps (or anything else that
// replaces the value object identity from the parent).
useEffect(() => {
setDraftPrompt(value.prompt || '');
setDraftNeg(value.negativePrompt || '');
}, [value.presetId, value.prompt, value.negativePrompt]);
const pickPreset = (presetId) => {
if (presetId === STYLE_ID.NONE) {
onChange?.(EMPTY_IMAGE_STYLE);
return;
}
if (presetId === STYLE_ID.CUSTOM) {
// Keep whatever's in the textarea — just flip the discriminator.
onChange?.({ presetId: STYLE_ID.CUSTOM, prompt: draftPrompt, negativePrompt: draftNeg });
return;
}
const preset = presets.find((p) => p.id === presetId);
if (!preset) return;
onChange?.({ presetId: preset.id, prompt: preset.prompt, negativePrompt: preset.negativePrompt });
};
const commitDraft = () => {
if (draftPrompt === value.prompt && draftNeg === value.negativePrompt) return;
// If the user edited a preset's text, flip to 'custom' so the dropdown
// reflects that the prompt no longer matches the curated preset.
const matchingPreset = presets.find((p) => p.id === value.presetId);
const stillMatchesPreset = matchingPreset
&& matchingPreset.prompt === draftPrompt
&& matchingPreset.negativePrompt === draftNeg;
onChange?.({
presetId: stillMatchesPreset ? value.presetId : STYLE_ID.CUSTOM,
prompt: draftPrompt,
negativePrompt: draftNeg,
});
};
return (
<div className="space-y-1.5">
<div className="flex items-center justify-between">
<span className="text-[9px] uppercase tracking-wider text-gray-500 flex items-center gap-1">
<Palette size={10} /> World style
</span>
{value.presetId !== STYLE_ID.NONE && (
// 44px tap target; -my-1 stays inside the 1.5 stack gap so the box
// never overlaps the preset <select> below it.
<button
type="button"
onClick={() => pickPreset(STYLE_ID.NONE)}
className="shrink-0 flex items-center justify-center min-w-[44px] min-h-[44px] -my-1 text-[9px] text-gray-500 hover:text-port-error"
title="Clear style"
>
Clear
</button>
)}
</div>
<select
aria-label="Preset"
value={value.presetId}
onChange={(e) => pickPreset(e.target.value)}
className="w-full bg-port-bg border border-port-border rounded px-2 py-1 text-[11px] text-gray-200"
>
<option value={STYLE_ID.NONE}>None — use scene visualPrompt only</option>
{groupPresetsByCategory(presets).map(([cat, items]) => (
<optgroup key={cat} label={cat}>
{items.map((p) => (
<option key={p.id} value={p.id}>{p.label}</option>
))}
</optgroup>
))}
<option value={STYLE_ID.CUSTOM}>Custom</option>
</select>
{value.presetId !== STYLE_ID.NONE && (
<>
<label htmlFor="wr-style-prompt" className="block">
<span className="text-[9px] uppercase tracking-wider text-gray-500">
Style prompt {value.presetId === STYLE_ID.CUSTOM && <Check size={9} className="inline text-port-accent" />}
</span>
<AutoSizeTextarea
id="wr-style-prompt"
value={draftPrompt}
onChange={(e) => setDraftPrompt(e.target.value)}
onBlur={commitDraft}
rows={3}
placeholder="cinematic still, anamorphic lens…"
className="w-full mt-0.5 bg-port-bg border border-port-border rounded px-2 py-1 text-[11px] text-gray-200 font-sans min-h-[60px]"
/>
</label>
<label htmlFor="wr-neg-prompt" className="block">
<span className="text-[9px] uppercase tracking-wider text-gray-500">Negative prompt (optional)</span>
<AutoSizeTextarea
id="wr-neg-prompt"
value={draftNeg}
onChange={(e) => setDraftNeg(e.target.value)}
onBlur={commitDraft}
rows={2}
placeholder="cartoon, low quality…"
className="w-full mt-0.5 bg-port-bg border border-port-border rounded px-2 py-1 text-[11px] text-gray-200 font-sans min-h-[44px]"
/>
</label>
<div className="text-[10px] text-gray-500">
Style is prepended to every scene's image prompt. Re-render scenes to see the change.
</div>
</>
)}
</div>
);
}
const RES_PRESETS = [
{ label: '768×512 (3:2)', width: 768, height: 512 },
{ label: '512×512 (1:1)', width: 512, height: 512 },
{ label: '512×768 (2:3)', width: 512, height: 768 },
{ label: '1024×576 (16:9)', width: 1024, height: 576 },
{ label: '1024×1024 (1:1)', width: 1024, height: 1024 },
];
// Per-mode form layout — Codex picks model/steps/seed internally, External
// SD-API loads its model server-side. Showing fields the active backend
// can't act on misleads the user about what's actually controlling renders.
const MODE_HINT = {
[IMAGE_GEN_MODE.LOCAL]: 'Renders locally via mflux/diffusers on this machine.',
[IMAGE_GEN_MODE.CODEX]: "Codex's $imagegen skill renders via your logged-in Codex session. Model, steps, and seed are picked by Codex itself.",
[IMAGE_GEN_MODE.EXTERNAL]: 'External SD-API renders against your configured A1111/Forge endpoint. The active model is set by the SD-API server.',
};
function ImageGenSettingsRow({ cfg, models, availableBackends = [], onChange }) {
const presetMatch = RES_PRESETS.find((p) => p.width === cfg.width && p.height === cfg.height);
const currentModel = models.find((m) => m.id === cfg.modelId);
const inputCls = 'w-full mt-0.5 bg-port-bg border border-port-border rounded px-2 py-1 text-[11px] text-gray-200 focus:border-port-accent outline-none';
const labelCls = 'text-[9px] uppercase tracking-wider text-gray-500';
const isLocalMode = cfg.mode === IMAGE_GEN_MODE.LOCAL;
const isCodexMode = cfg.mode === IMAGE_GEN_MODE.CODEX;
return (
<div className="space-y-1.5">
{availableBackends.length > 0 && (
<div>
<span className={labelCls}>Backend</span>
<div className="mt-0.5">
<BackendChipStrip
availableBackends={availableBackends}
value={cfg.mode}
onChange={(id) => onChange({ ...cfg, mode: id })}
size="sm"
ariaLabel="Image gen backend"
titlePrefix="Render storyboard scenes via"
/>
</div>
<p className="text-[9px] text-gray-500 mt-1">{MODE_HINT[cfg.mode]}</p>
</div>
)}
<div className="grid grid-cols-2 gap-1.5">
{isLocalMode && (
<label className="block">
<span className={labelCls}>Image model</span>
<select
value={cfg.modelId}
onChange={(e) => onChange({ ...cfg, modelId: e.target.value })}
className={inputCls}
>
{models.length === 0 && <option value={cfg.modelId}>{cfg.modelId}</option>}
{models.map((m) => (
<option key={m.id} value={m.id}>{m.name || m.id}</option>
))}
</select>
</label>
)}
<label className="block">
<span className={labelCls}>Resolution</span>
<select
value={presetMatch ? `${cfg.width}x${cfg.height}` : 'custom'}
onChange={(e) => {
if (e.target.value === 'custom') return;
const [w, h] = e.target.value.split('x').map(Number);
onChange({ ...cfg, width: w, height: h });
}}
className={inputCls}
>
{RES_PRESETS.map((p) => (
<option key={p.label} value={`${p.width}x${p.height}`}>{p.label}</option>
))}
{!presetMatch && <option value="custom">Custom ({cfg.width}×{cfg.height})</option>}
</select>
</label>
</div>
{!isCodexMode && (
<div className="grid grid-cols-2 gap-1.5">
<label className="block">
<span className={labelCls}>
Steps {isLocalMode && currentModel?.steps && <span className="normal-case text-gray-600">(default {currentModel.steps})</span>}
</span>
<input
type="number" min={1} max={150}
value={cfg.steps}
onChange={(e) => onChange({ ...cfg, steps: e.target.value })}
placeholder={isLocalMode ? String(currentModel?.steps || 'auto') : 'server default'}
className={inputCls}
/>
</label>
<label className="block">
<span className={labelCls}>Seed</span>
<div className="flex items-stretch gap-1 mt-0.5">
<input
type="number"
value={cfg.seed}
onChange={(e) => onChange({ ...cfg, seed: e.target.value })}
placeholder="random"
className="flex-1 bg-port-bg border border-port-border rounded px-2 py-1 text-[11px] text-gray-200 focus:border-port-accent outline-none"
/>
<button
type="button"
onClick={() => onChange({ ...cfg, seed: randomSeed() })}
className="min-h-[44px] min-w-[44px] flex items-center justify-center text-gray-500 hover:text-port-accent border border-port-border rounded"
title="Randomize seed"
aria-label="Randomize seed"
>
<Dice5 size={11} />
</button>
</div>
</label>
</div>
)}
</div>
);
}
// ─── Config (image gen + style + Adapt LLM) ───────────────────────────────
export default function StoryboardConfigTab({ imageCfg, models, availableBackends, onCfgChange, stylePresets, imageStyle, onStyleChange, onOpenImageGenSettings }) {
return (
<div className="px-3 py-3 space-y-4">
<section className="space-y-1.5">
<div className="text-[12px] font-semibold text-gray-200">Adapt LLM</div>
<StagePromptModelPicker
stageName={SCRIPT_STAGE}
label="Adapt LLM"
icon={<Cpu size={10} />}
hint="Used when you click Run Adapt to break prose into scenes."
/>
</section>
<section className="space-y-1.5 pt-3 border-t border-port-border">
<div className="text-[12px] font-semibold text-gray-200">World style</div>
<WorldStyleRow value={imageStyle} presets={stylePresets} onChange={onStyleChange} />
</section>
<section className="space-y-1.5 pt-3 border-t border-port-border">
<div className="flex items-center justify-between gap-2">
<div className="text-[12px] font-semibold text-gray-200">Image generation</div>
<button
type="button"
onClick={onOpenImageGenSettings}
className="flex items-center gap-1 px-2 py-1 text-[10px] rounded border border-port-border text-gray-300 hover:bg-port-border/40 hover:text-white"
title="Configure image gen backends — enable Codex $imagegen, set local Python, etc."
>
<SettingsIcon size={10} /> Backends
</button>
</div>
{availableBackends.length === 0 && (
<div className="text-[10px] text-port-warning bg-port-warning/10 border border-port-warning/40 rounded px-2 py-1.5">
No image gen backend configured. Click <span className="font-medium">Backends</span> to enable Local mflux, Codex <code className="text-gray-400">$imagegen</code>, or an External SD-API endpoint.
</div>
)}
<ImageGenSettingsRow cfg={imageCfg} models={models} availableBackends={availableBackends} onChange={onCfgChange} />
</section>
</div>
);
}