Essentials core · v0.1.0
Clock, date, notes and pictures.
core/clockBig digital clock, with optional seconds and date.
core/dateDay of the month, weekday and week number, like a desk calendar.
core/textA note or a motto, auto-sized to fill its tile.
core/imageA local image or a URL (refreshed every few minutes, good for webcams and charts).aimonitor compatibility legacy · v0.1.0
Run your aimonitor custom modules as widgets.
legacy/moduleRuns a custom card written for aimonitor (defineModule) unchanged.AI agents agents · v0.1.0
Claude Code and Gemini CLI sessions at a glance: who waits for you, for how long, and git status. Ported from aimonitor.
agents/sessionsOne tile per AI coding session. The one that needs you most turns solid, the others that wait get a colour spine, and working ones stay dark.
agents/summaryA thin header line: RAM, how many consoles are open, how many wait for you (tinted by the most urgent) and the clock.Cats room cats · v0.1.0
Your AI agent sessions as cats in a room: what needs you comes forward. From aimonitor.
cats/roomEach AI agent session is a cat in a room with the real sky outside: the ones that need you come forward, the ones working sit at their laptops in the back.System system · v0.1.0
Machine vitals as a shared source, and a system health card.
system/healthDisk usage, CPU temperature over the last hour, load, memory, uptime and failed systemd units of this computer.Vitals vitals · v0.1.0
The machine at a glance, quiet until something runs hot or fills up. Needs the system vitals source.
vitals/vitalsCPU temperature, CPU, RAM and disk in dim ink that only turns amber or red past your thresholds; per-core bars when wide.AI usage usage · v0.1.0
How much Claude Code you are burning, from ccusage: limit windows or spend, tokens by model, spend per block.
usage/railLimit windows (or spend against your caps) as bars with their reset time, and today’s tokens by model.
usage/blocksThe equivalent cost of the last twelve 5-hour blocks, the running one highlighted: is today unusual?Weather weather · v0.1.0
Weather from Open-Meteo, shared with every widget, and a forecast card.
weather/forecastTemperature and sky now, the next hours (temperature curve and rain) and days, and the moon when there is room.Calendar calendar · v0.1.0
Meetings from published ICS calendars (Outlook/Teams, Google…): a countdown to the next one and the agenda of today and tomorrow.
calendar/nextA big countdown to your next meeting. Its colour climbs the semaphore as it nears, and it takes the tile over when it’s about to start.
calendar/agendaToday’s remaining meetings and tomorrow’s: the one under way highlighted, the next one with its countdown.News news · v0.1.0
Press headlines from RSS feeds with their photos. No keys and no AI: consensus across outlets measured with plain set overlap.
news/headlineOne story at a time over its photo, full-bleed: outlet, age, how many outlets carry it, headline and standfirst. With two versions of the same fact side by side when outlets tell it differently.
news/columnThe current story in a column for a shared screen: photo with rounded corners, clock and temperature on top, headline and standfirst below.Mirror cards mirror · v0.1.0
MagicMirror²-style cards: countdown, compliments, today and tomorrow, now playing, on this day, air quality, earthquakes and rain radar.
mirror/countdownDays left until your important dates, nearest first.
mirror/complimentsA phrase that suits the time of day, like the MagicMirror² classic.
mirror/tasksYour to-dos for today and tomorrow, from a Markdown checklist and, optionally, your GitLab issues.
mirror/nowplayingThe song playing on this computer (any MPRIS player on Linux: Spotify, browser, VLC…), with cover and progress. Hidden while nothing plays.
mirror/onthisdaySelected anniversaries of today from Wikipedia, one at a time.
mirror/airqualityEuropean air quality index, particles, gases and pollen where you are (Open-Meteo).
mirror/earthquakesRecent earthquakes around you (EMSC) on a polar map, with magnitude, place and time. Hidden when there are none.
mirror/rainmapLatest rain radar (RainViewer) over a dark OpenStreetMap map, home in the centre. By default only shows while it rains nearby.GitLab gitlab · v0.1.0
Your merge requests' pipelines and the reviews waiting for you.
gitlab/pipelinesThe pipeline of each of your open merge requests: broken first, then running and queued, passing last.
gitlab/reviewsMerge requests waiting for your review: whose, how long they have waited and how their pipeline is doing. The ones you approved sink, dimmed.GitHub github · v0.1.0
Your Actions runs, the pull requests waiting for your review and your own.
github/actionsRecent GitHub Actions workflow runs: broken first, then running (with a timer) and queued, passing last.
github/reviewsPull requests waiting for your review (whose, how long they have waited, how their checks are doing), then your own with what blocks them: failing checks, changes requested, approved. Mentions last, if you want them.
github/inboxThree numbers: reviews waiting for you, failing runs and unread notifications. Coloured only when they ask for you.Calls calls · v0.1.0
Detects video calls by the open microphone, even muted, and shows an on-air light.
calls/statusAn on-air light: lit while an app holds the microphone (even muted), with how long the call has lasted. Dark otherwise.Timesheet timesheet · v0.1.0
Hours logged in GitLab, yours and your group's, and what is still left to log today.
timesheet/hoursHours logged in GitLab by you and your group: today, this week, daily and weekly averages, and each one's trend against themselves. Yours turn green or red against what is due by now.
timesheet/focusWhat you have logged today against what is due by now, and what is still unlogged: finished meetings and agent activity per project.Local LLMs llm · v0.1.0
Tokens per second, loaded models and request queues of your local LLM servers.
llm/throughputGeneration tokens per second of each server in big figures, prompt processing in small, and a sparkline of the last minutes. Dim while idle.
llm/modelsThe models in memory across your servers: size and VRAM, context length and, for Ollama, how long until keep-alive unloads them.
llm/queueRequests running and waiting on each server, with its KV cache usage. Turns red when too many wait or the cache is nearly full.
llm/statusOne line per server for small panels: up or down, the loaded model and its tokens per second.GPU gpu · v0.1.0
The GPUs of your AI box: temperature, load, VRAM by process and history. Local, over ssh or from a Prometheus exporter.
gpu/cardsOne column per GPU: temperature, load, VRAM and power bars, fan and throttling. Dim until a threshold is crossed.
gpu/vramVRAM per GPU as a bar stacked by process (ollama, vllm, python…), with what is left free.
gpu/historyTemperature line over load area for the last minutes, per GPU, with the critical line dashed.