RisalDash

Beautiful real-time web dashboards for ESP32 / ESP8266 — in a few lines of C++.

Describe widgets; RisalDash generates the HTML, CSS, JS and the WebSocket protocol for you. The dashboard is served by the device itself in an OKLCH "liquid glass" style (translucent cards, iOS-like status bar, a Settings gear for language/theme/accent, swipe-up multi-page layouts), updates live over WebSocket, and works offline-first — including a captive portal for first-boot Wi-Fi setup. Zero front-end code.

🌐 dash.risal.io · MIT · ESP32 + ESP8266

#include <RisalUI.h>
RisalUI dash("Greenhouse");

float temp = 24.3, volts = 12.1; int bright = 128; bool pump = false;

void setup() {
  dash.gauge ("Voltage", &volts, 0, 14, "V");
  dash.chart ("Temperature", &temp, "C");
  dash.slider("Brightness", &bright, 0, 255, [](int v){ analogWrite(LED_PIN, v); });
  dash.toggle("Pump", &pump, [](bool on){ digitalWrite(PUMP_PIN, on); });
  dash.begin();        // saved Wi-Fi → connect; first boot → captive setup portal
}

void loop() {
  temp = readTemp(); volts = readVolts();
  dash.update();       // pushes changed values to the browser
}

Why RisalDash

• Zero-Waste UI — the linker (--gc-sections) strips widget types you don't use (0 bytes). A type you do use adds its own C++ + CSS + JS once (~1.3–3.4 KB, measured on ESP32).
• Offline-first first boot — begin() raises a Wi-Fi access point with a captive portal; the user picks a network and the credentials are saved to NVS. No internet, no app, no CDN (system fonts, everything served from flash).
• Real-time — values are pushed over WebSocket only when they change; controls send commands back to your callbacks.
• Widgets for everything — 26 types: displays, controls, layout (tabs/groups/span), plus one-line sensor presets.
• Multi-page + native chrome — dash.layout() splits the UI into pages switched by a swipe-up sheet of icon tiles; an iOS-style status bar (clock, Wi-Fi, battery) sits on top.
• Settings on the device — a gear in the appbar opens Language / Theme / Accent, applied live and remembered per browser. dash.lang("en"|"ru"|"ar") sets the default; Arabic flips to RTL.
• Integrations — REST, Prometheus /metrics, optional MQTT, OTA firmware update, and MCP so an AI agent can read sensors and drive controls (every widget becomes a tool).
• Brand-consistent — the same OKLCH design system as the app and dash.risal.io.

Before / After

A browser UI for an ESP usually means hand-writing HTML + CSS + JS and a WebSocket protocol — easily 100+ lines for a couple of gauges. RisalDash is just the declaration.

server.on("/", HTTP_GET, [](AsyncWebServerRequest* r) {
  r->send(200, "text/html", R"HTML(
    <div id="t">--</div><label><input type="checkbox" id="p"> Pump</label>
    <script>
      let ws; (function c(){ ws = new WebSocket('ws://'+location.host+'/ws');
        ws.onmessage = e => { const s = JSON.parse(e.data); t.textContent = s.temp; p.checked = s.pump; };
        ws.onclose = () => setTimeout(c, 800); })();
      p.onchange = () => ws.send(JSON.stringify({ pump: p.checked }));
    </script>
    <style>/* gauge SVG, layout, theme, fonts, mobile… */</style>
  )HTML");
});
ws.onEvent([](/*…*/ AwsEventType ty, uint8_t* d, size_t n) {
  if (ty == WS_EVT_DATA) { /* parse JSON, find the key, apply to your var, call your callback */ }
});
void loop() {
  if (millis() - last > 250) {                       // throttle by hand
    String j = "{"; j += "\"temp\":" + String(temp) + ",\"pump\":" + (pump ? "true" : "false") + "}";
    ws.textAll(j); last = millis();                  // build + broadcast JSON by hand
  }
  // …now repeat all of this for every new widget, plus the CSS and the protocol.
}

After — RisalDash:

dash.gauge ("Temperature", &temp, 0, 50, "C");
dash.toggle("Pump", &pump, [](bool on){ digitalWrite(PUMP, on); });
dash.begin();                                        // + offline captive portal, i18n, OTA, MCP…

Install

Arduino IDE — Library Manager → search "RisalDash".

PlatformIO — platformio.ini:

lib_deps =
    RisalDash
    esp32async/ESPAsyncWebServer
    esp32async/AsyncTCP        ; ESP32
  ; esp32async/ESPAsyncTCP    ; ESP8266

ESP-IDF (Arduino as a component) — from the Component Registry:

idf.py add-dependency "shaxzodahmedov/risaldash"

Wi-Fi: first boot vs. fixed credentials

dash.begin();                          // saved creds → STA; otherwise captive setup portal
dash.begin("ssid", "password");        // connect to this network (falls back to the portal)
dash.beginAP("Greenhouse", "12345678");// plain dashboard over its own access point
dash.apName("Greenhouse-Setup");       // name of the captive-portal AP (optional)

On first boot the device appears as a RisalDash-Setup Wi-Fi. Connect to it — the setup page opens automatically (captive portal). Pick your network, enter the password; the device reboots and serves the dashboard on your Wi-Fi.

Widgets

All widgets bind to a variable by pointer and update live.

Layout: group(title), separator(title), tab(title) (switchable panels).

Sizes: every widget sits on an iOS-style cell grid and has a sensible default footprint. Override it with .size(RSIZE_S) (1 cell, compact), .size(RSIZE_M) (full width) or .size(RSIZE_L) (full width + double height) — e.g. dash.metric("CPU", &cpu, "%").size(RSIZE_L);. Works on single widgets and on sensor presets: dash.sensor("bme280", …).size(RSIZE_L).

Icons: .icon(RICON_THERMOMETER) puts an IoT glyph in the card header. Built-in set: thermometer, water, flash, bulb, power, gauge, home, wifi, clock, signal, leaf, motion — or pass any 24×24 SVG path. Only the icons you use are linked into flash.

Pages, status bar & appearance

Split the dashboard into pages — each dash.layout() starts one; the widgets after it belong to that page, and a swipe-up sheet of icon tiles (or the bottom handle) switches pages:

dash.layout("Overview", RICON_HOME);
dash.metric("CPU", &cpu, "%");
dash.layout("Climate", RICON_THERMOMETER);
dash.slider("Target", &target, 16, 30);

Every page carries an iOS-style status bar (clock, Wi-Fi, battery) and an appbar Settings gear (Language / Theme / Accent, remembered per browser). Set the defaults from the sketch:

dash.timezone(180);          // status-bar clock & portal default, minutes from UTC (+03:00)
dash.accent(2);              // 0 Aqua · 1 Blue · 2 Violet · 3 Amber · 4 Rose
dash.theme(RisalUI::DARK);   // DARK (default) | LIGHT | AUTO

Sensor presets

One line drops the right widgets, units and ranges for a known sensor:

dash.sensor("bme280", &temp, &hum, &pres);  // gauge °C + metric % + chart hPa
dash.sensor("ina219", &volts, &cur, &pwr);  // V / A / W

Built-in: bme280, bmp280, dht11, dht22, sht3x, ds18b20, bh1750, ccs811, scd40, ina219, acs712, pzem004t, hcsr04, vl53l0x, mq135, soil, ld2410, ld2450, mpu6050, mpu9250, neo-m10, inmp441. The widget is chosen by the quantity, not the sensor model.

Fake sensors — build with no hardware

Prototype and debug the whole dashboard with nothing plugged in. #include <RisalFake.h> gives you realistic drifting readings — a slow trend + wobble + a little noise, not a flat sine — so the UI looks live while you iterate on layout and logic.

#include <RisalFake.h>
RisalFakeEnv env;                 // temperature / humidity / pressure / soil / air quality

void setup() { env.begin(); /* ...declare widgets bound to temp/hum/... ... */ }
void loop() {
  env.update();
  temp = env.temperature();       // °C   — later: aht.readTemperature()
  hum  = env.humidity();          // %
  pres = env.pressure();          // hPa
  dash.update();
}

Need one value of any other quantity? RisalFake(center, amp, noise):

RisalFake volts(12.4, 0.3, 0.05);   // 12.4 V, ±0.3 drift, a little jitter
voltage = volts.read();

When the real sensor arrives, swap the reads for your driver's — same variable names, the rest of the sketch is unchanged. Opt-in and Zero-Waste: nothing is compiled unless you include it. Works on ESP8266 and ESP32. The ESP32-C6-LCD-1.47 example ships with it.

Languages

dash.lang("ar");   // default: "en" | "ru" | "ar"  — "ar" switches to RTL

The appbar Settings gear lets the user switch language (EN / RU / AR) live too. Only the languages you reference are compiled in (Zero-Waste); widget titles stay yours, the library chrome is translated.

Integrations & control

dash.enableMCP("risal_pat_token");   // GET /api/mcp/manifest → AI tools (see tools/risal-mcp-bridge)
dash.enableOTA();                    // /update page + "Update firmware" button in Settings (OTA)
dash.mqtt("broker.local", 1883, "greenhouse");  // needs -D RISAL_ENABLE_MQTT + PubSubClient
dash.enableHomeAssistant("greenhouse");          // Home Assistant MQTT auto-discovery (after mqtt())

MCP — enableMCP(token) exposes GET /api/mcp/manifest, turning every widget into an AI tool (read sensors, drive controls). The companion risal-dash-mcp bridge connects a device to Claude Desktop / Claude Code. 📝 Walkthrough: Control your ESP32 from an AI agent.

OTA — enableOTA() serves /update and adds an "Update firmware" button to the Settings modal, so you flash a new .bin from the browser — no USB. It's opt-in on purpose: OTA needs a partition with a second app slot (Arduino's default "4MB with spiffs" and the ESP8266 4MB layouts have one; huge_app / single-app schemes do not), and /update lets anyone on the LAN flash the device — so turn it on for a trusted network. Every dashboard also has a built-in Reset Wi-Fi button in Settings (forgetWiFi() → back to the setup portal).

Home Assistant — after mqtt(), enableHomeAssistant() publishes MQTT discovery configs so HA auto-creates entities (sensors, switches, numbers, binary sensors, buttons), all grouped under one device. No YAML.

Examples

• Minimal — a few widgets over an access point.
• FirstBoot — captive-portal Wi-Fi provisioning (signal levels, timezone), then your network.
• WifiSetup — the full provisioning circle, incl. dash.forgetWiFi() back to the portal.
• Layouts — multi-page dashboard with the swipe-up page switcher + accent()/timezone().
• Tabs — switchable tab panels, with cards pinned above them.
• AllWidgets — every widget type across five pages, from gauges to the robot face, map, 3D cube, terminal and thermal heatmap.
• WidgetSizes — the cell grid: the same reading at RSIZE_S / RSIZE_M / RSIZE_L, compact gauges, and a sensor preset resized as a group.
• FakeSensors — the whole <RisalFake.h> toolbox: drifting values, the day/night greenhouse, GPS route playback, a fake BLE scan and record-&-replay — no hardware attached.
• SensorTemplates — dash.sensor("bme280", …) presets, tuned with .chart() / .size().
• UiKit — the interface skeleton: pages, groups, separators, sizes, icons, .gear(), chrome.
• Multilingual — dash.translate(): the whole UI (your titles included) switches EN/RU/UZ/AR.
• ESP32-C6-LCD-1.47 — the big showcase: web dashboard + a 13-slide LCD carousel, real BLE scan, live weather, robot eyes; split into display.h / led.h / i18n.h / sensors.h / weather.h / ble.h so the sketch stays readable.
• ESP32-S3-DualCore — a heavy worker on core 0 while the dashboard stays smooth on core 1.
• Wemos-D1-Mini — all-real-hardware ESP8266 showcase: onboard LED, A0, live heap/RSSI.

Footprint

RisalDash is Zero-Waste: a widget type you never call is stripped by the linker, so it costs nothing. Measured on ESP32 (Arduino core 3.3) — flash added by the first use of each type (its C++ + CSS + JS), over an empty RisalDash dashboard:

Extra instances of a type you already use are a few bytes each. Unused types: 0 bytes — that's the point.

Roadmap

Richer charts (multi-series / area / bar), more sensor presets, CSS/JS minify + gzip-in-PROGMEM, Home Assistant auto-discovery, and a Wokwi simulation link. See dash.risal.io.

License

MIT © ZiyaraGo Technologies. Brand: Risal.