The all-in-one local AI studio for your desktop: chat, image and video generation and a coding agent in one free, open source app. Windows and Linux. No Docker, no terminal, no cloud required.
TypeScript agentai-chatai-studioall-in-onecoding-agentdesktop-appfluximage-generationlinuxllmlocal-ailocal-llmopen-sourceprivacyself-hostedstable-diffusionuncensoreduncensored-aivideo-generationwindows
import { execSync } from 'child_process'
import { statfsSync } from 'fs'
import os from 'os'
import type { RouteMount } from './routes'
import { errorText } from '../src/types/json-guards'
/** Die vier Auskunfts-Endpunkte über den Rechner. Sie lesen nur. */
export function registerSystemRoutes(routes: RouteMount): void {
// API: System info
routes.use('/local-api/system-info', (_req, res) => {
res.writeHead(200, { 'Content-Type': 'application/json' })
res.end(JSON.stringify({
os: process.platform, arch: process.arch, hostname: os.hostname(),
username: os.userInfo().username, totalMemory: os.totalmem(), cpuCount: os.cpus().length,
}))
})
// API: System health (mirrors the Rust `system_health` command so the
// Settings → Troubleshoot "Re-probe" button works under `npm run dev`
// too. The plain dev server previously had no /local-api/system-health,
// so the button errored (2026-06-07). Dev-only — the
// packaged app uses the real Rust probe.
routes.use('/local-api/system-health', async (_req, res) => {
const probe = async (url: string, endpoint: string) => {
try {
const r = await fetch(url)
return { status: r.ok ? 'ok' : 'error', detail: `HTTP ${r.status}`, endpoint }
} catch (e) {
return { status: 'unreachable', detail: errorText(e) || String(e), endpoint }
}
}
const ollamaBase = (process.env.OLLAMA_HOST && /^https?:/.test(process.env.OLLAMA_HOST))
? process.env.OLLAMA_HOST.replace(/\/+$/, '')
: 'http://localhost:11434'
const [ollama, comfyui, lm_studio] = await Promise.all([
probe(`${ollamaBase}/api/tags`, ollamaBase),
probe('http://localhost:8188/system_stats', 'http://localhost:8188'),
probe('http://localhost:1234/v1/models', 'http://localhost:1234'),
])
let vram_total_gb: number | null = null
let vram_free_gb: number | null = null
try {
const out = execSync('nvidia-smi --query-gpu=memory.total,memory.free --format=csv,noheader,nounits',
{ encoding: 'utf8', timeout: 4000 })
const [tot, free] = String(out).trim().split('\n')[0].split(',').map((s: string) => parseFloat(s.trim()))
if (!isNaN(tot)) vram_total_gb = +(tot / 1024).toFixed(1)
if (!isNaN(free)) vram_free_gb = +(free / 1024).toFixed(1)
} catch { /* no nvidia-smi → null */ }
let disk_free_gb = 0
try {
const st = statfsSync(os.homedir())
disk_free_gb = +((Number(st.bavail) * Number(st.bsize)) / 1e9).toFixed(1)
} catch { /* statfs unavailable */ }
res.writeHead(200, { 'Content-Type': 'application/json' })
res.end(JSON.stringify({
version: 'dev',
host: {
os: process.platform, os_version: os.release(), arch: process.arch,
cpu_count: os.cpus().length, ram_gb: +(os.totalmem() / 1e9).toFixed(1),
disk_free_gb, vram_total_gb, vram_free_gb,
},
ollama, comfyui, lm_studio,
}))
})
// API: Process list
routes.use('/local-api/process-list', (_req, res) => {
// Simple stub — full process list needs native code
res.writeHead(200, { 'Content-Type': 'application/json' })
res.end(JSON.stringify({ processes: [], count: 0, note: 'Full process list available in Tauri build only' }))
})
// API: Screenshot
routes.use('/local-api/screenshot', (_req, res) => {
res.writeHead(200, { 'Content-Type': 'application/json' })
res.end(JSON.stringify({ error: 'Screenshot available in Tauri build only' }))
})
}