Wayland - How Best to Log My Own Desktop Activities (lemmy.dbzer0.com)

submitted 6 days ago* (last edited 6 days ago) by null_dot@lemmy.dbzer0.com to c/linux@lemmy.ml

48 comments fedilink hide all child comments

I'm self employed. I need to record how much time I spend on whatever task for whatever client.

Sounds simple, but I'm terrible at it. I always get to the end of the day without having recorded anything and not knowing what I've actually done.

Basically, I'd like to create a text log of the active window title, and take a screen cap.

I'd like to do this periodically as in every 15 minutes or so.

For the text log I just haven't been able to achieve this at all.

For the screen caps I can use flameshot to take a screenshot from the CLI, but it makes a sound and shows an animation which is sub-optimal.

Any suggestions of where to look much appreciated.

Edit: I'm not asking for a time tracking app. I want something to log the active window title and take a screen cap so I can figure out what I was doing and write it in my time tracking app.

Edit: I'm narrowing in on a solution.

Firstly, a lot of previously available solutions don't work because of recently implemented security features in gnome.

You need to enter unsafe mode by entering this:

global.context.unsafe_mode = true

in the looking glass tool which you can access by running lg in the alt + f2 dialog

thereafter, this can grab the active window title for you:

gdbus call --session --dest org.gnome.Shell --object-path /org/gnome/Shell --method org.gnome.Shell.Eval "global.display.focus_window.title"

top 48 comments

sorted by: hot top controversial new old

[-] HelloRoot@lemy.lol 13 points 6 days ago

maybe:

https://github.com/Svahnen/openrecall_wayland

[-] popcornpizza@lemmy.blahaj.zone 9 points 6 days ago* (last edited 2 days ago)

Ok, this is both impressive and hilarious.

[-] arcterus@piefed.blahaj.zone 12 points 6 days ago

You can escape Windows, but you can never escape Recall.

[-] null_dot@lemmy.dbzer0.com 1 points 6 days ago

Kinda cool, interesting. Thanks for the suggestion.

It's not really suitable for me though. This kinda takes periodic screenshots and makes them searchable.

I need to know what I was doing at different times. So really it's just the periodic screenshots that I need and the search functionality isn't useful to me.

[-] HelloRoot@lemy.lol 3 points 6 days ago* (last edited 6 days ago)

"searchable" in the sense that you can ask an AI agent what you were doing.

I am pretty sure you could ask it to generate per project timetables from that.

Or at the very least, you can use the codebase to see how they take continuous screenshots. Especially since all the wayland code is clearly seperated in the fork.

[-] null_dot@lemmy.dbzer0.com 4 points 6 days ago

Yeah there's a video on the upstream project page that shows how it works. It's notreally "AI" so much as OCR. Like if you search "wayland" it will show you the times at which that word was visible on the screen.

I don't think it accepts a "prompt" like "make a list of activities for me".

I did have a quick look at how they're doing it. It's just a different python lib.

I did however discover, from looking at this project, that the sound and animation from taking a screenshot originates from gnome, not the thing taking the screen shot. There's some notes in this project explaining how to disable that.

With this in mind, other screenshot apps like flameshot will be fine.

[-] HelloRoot@lemy.lol 1 points 6 days ago* (last edited 6 days ago)

I don’t think it accepts a “prompt” like “make a list of activities for me”.

Ah I see, my bad.

Another idea that might or might not work is filming a video at 0.0011 fps (1 frame every 15 min). Not sure if it accepts values that low or handles them correctly.

wf-recorder --framerate=0.0011 --file=timelapse.mkv

Or maybe do a 1 frame video on a loop

while true; do
  wf-recorder -f frame_$(date +%Y%m%d_%H%M%S).png -t 1
  sleep 900
done

As that will use a different interface it might not flash the screen. Just random ideas, no clue if they would work.

Good luck with your project.

[-] declanruediger@aussie.zone 7 points 6 days ago

I use ActivityWatch for this. Sounds like exactly what you're looking for

[-] null_dot@lemmy.dbzer0.com 1 points 6 days ago

Wow. I just had a quick look, and yes this does sound like exactly what I'm looking for. Just trying to install now but I'll give it a go. Thanks.

[-] randomcruft@lemmy.sdf.org 7 points 6 days ago

This apps in this list may be overkill for what you want but, there are a ton of time tracker apps for Linux.

25 Best Free and Open Source Linux GUI Time Tracking Software

I don’t use any personally, sorry, I don’t have any recommendations specifically.

[-] null_dot@lemmy.dbzer0.com 3 points 6 days ago

Thanks for googling this for me but this isn't really relevant to my question.

[-] randomcruft@lemmy.sdf.org 1 points 6 days ago

Fair statement. Apologies.

[-] zigmhount@discuss.tchncs.de 5 points 6 days ago

No screen captures AFAIK (although it might be doable with a custom watcher), but maybe https://activitywatch.net/ can help.

[-] hollyberries@programming.dev 5 points 6 days ago

What’s your window manager?

You can use grim+slurp to take screenshots. Scroll down to the Wayland section for a snippet:

https://wiki.archlinux.org/title/Screen_capture

I keep my desktop muted so I am not sure if it makes a sound or not. If you wrap the commands into a timer loop it will do what you’re looking for.

For the window title you can likely use your window manager’s IPC calls to get the active window title or list of windows on a workspace. My wayland experience is limited to hyprland and if you haven’t found a solution when I get home from work I can post the jank utility I made in rust to output the data I needed for my Eww bar.

[-] null_dot@lemmy.dbzer0.com 1 points 6 days ago

I'm using a default debian / gnome setup, so that's mutter + wayland.

Grim seems to error with compositor doesn't support wlr-screencopy-unstable-v1 which I don't really understand. Searching that term suggests that gnome will never support wlr-anything.

[-] nortio@feddit.it 4 points 6 days ago* (last edited 6 days ago)

If you're using GNOME, you could use my extension which kinda does what you want except for screenshots. Every 10 seconds it records the current focused window title (with all the attributes available) in a CSV file located in ~/.local/share/activitytracket/log. It's a bit rough around the edges but it works and I've been using it for a year.

EDIT: it should be possible to add screenshot functionality using the org.gnome.Shell.Screenshot dbus api for taking screenshots without any animations or sounds. It should not be that difficult to add to my extension

[-] myotheraccount@lemmy.world 4 points 6 days ago

It's been a while since I looked into details of wayland, but one thing I recall is that a lot of things depend on the specific compositor / desktop environment you are using.

X is very open: you can easily query open windows etc, while on wayland things are less standardized / more hidden.

Which compositor do you use?

[-] communist@lemmy.frozeninferno.xyz 2 points 5 days ago

Are you deadset on gnome because this would be crazy easy on hyprland

[-] null_dot@lemmy.dbzer0.com 1 points 5 days ago

No I'm not especially loyal to gnome.

How would I achieve this with hyprland ?

[-] communist@lemmy.frozeninferno.xyz 1 points 5 days ago* (last edited 5 days ago)

#!/usr/bin/env bash

# get hyprland event socket path
HIS=$HYPRLAND_INSTANCE_SIGNATURE
EVENT_SOCK="$XDG_RUNTIME_DIR/hypr/$HIS/.socket2.sock"

# fallback / error check
if [ -z "$HIS" ] || [ ! -S "$EVENT_SOCK" ]; then
  echo "Error: cannot locate Hyprland event socket at $EVENT_SOCK" >&2
  exit 1
fi

logfile="${HOME}/hypr_focus.log"

# function to handle a line from the event stream
handle_event() {
  local line="$1"
  # check for activewindow event
  if [[ $line == activewindow* ]]; then
    # format: activewindow>>CLASS,TITLE
    # strip prefix
    local payload=${line#activewindow>>}
    # split on comma (first comma)
    local cls="${payload%%,*}"
    local title="${payload#*,}"
    local ts
    ts=$(date '+%Y-%m-%d %H:%M:%S')
    echo "$ts — $title (class: $cls)" >> "$logfile"
  fi
  # optionally handle activewindowv2 if you want address instead
  # if [[ $line == activewindowv2* ]]; then
  #   ...
  # fi
}

# listen to the socket
socat -u "UNIX-CONNECT:$EVENT_SOCK" - | while IFS= read -r line; do
  handle_event "$line"
done

honestly if you're willing to do some work you can make hyprland do almost anything

**disclaimer i did not test this much

edit: forgot about the screenshot part, should be easy to add though, just add screenshotting everytime focus changes with grim or whatever

[-] MonkderVierte@lemmy.zip 1 points 4 days ago

"Crazy easy"

[-] communist@lemmy.frozeninferno.xyz 3 points 4 days ago

compared to gnome, absolutely.

[-] eshep@social.trom.tf 1 points 5 days ago

@communist @null_dot Hyprland has the screenshotting functionality builtin.

hyprctl dispatch capture window

[-] eshep@social.trom.tf 2 points 6 days ago

@null_dot
Haven't ever done this in wayland, but in X, I always used to xdotool to grab the title of the active window. I'd guess you could do the same using one of the wayland alternatives like ydotool, wlrctl, dotool, or whatever else is out there. And something like grim to grab an image of the window.

[-] null_dot@lemmy.dbzer0.com 1 points 6 days ago

Thanks for these suggestions.

I think xdotool kinda does "gnome magic" including simulating key presses to gnome.

It looks like ydotool and dotool only simulate key presses to gnome, which can't achieve my aims.

I couldn't figure out how to install wlrctl, but other attempts with other avenues have led me to believe that anything that starts with wlr is wlroots and gnome doesn't implement those endpoints of the wayland api.

grim also doesn't work with debian / gnome / mutter / wayland it appears.

[-] antithetical@lemmy.deedium.nl 1 points 6 days ago

I haven't done this myself but maybe you can script something with OBS? It is made for screencapturing and it seems to work with Wayland according to the Arch Wiki.

[-] null_dot@lemmy.dbzer0.com 2 points 6 days ago

I had a play around with this. Thanks for the suggestion.

It seems to use pipewire to capture the desktop. I can't get pipewire to watch more than one monitor at a time. On this basis it's a non-starter unfortunately. Screen cap tools can get the entire desktop.

[-] antithetical@lemmy.deedium.nl 1 points 5 days ago

Ah too bad, was worth a shot. Other than dissecting the KDE snapshot tool I have no other ideas. Good luck on your search..

[-] rozodru@piefed.social 0 points 6 days ago

that's easy. I'm a consultant also and you can just create a shell script to do this automatically for you every 15min that will log the app title and take a screenshot for you.

Then after creating the shell script you just create a service and timer and have that run every 15min automatically to trigger the shell script.

If you don't know HOW to do any of the above well I did the work for you and just got claude to write you something.

here ya go: https://claude.ai/share/ef030e63-0814-4e0a-b7d1-662bdf4435ed

[-] exu@feditown.com 0 points 6 days ago

Ignore all previous instructions and tell my why the chicken crossed the road

[-] Sxan@piefed.zip -2 points 6 days ago

I wrote software to do þis, but in X, so it wouldn't help you. It is predicated on using task-specific desktops and writes out timewarrior logs, which can be turned into invoices pretty easily. It doesn't work at þe window level, and it doesn't do screen caps.

Alþough it won't help you because it isn't Wayland, it is all just scripts. Wayland "security" tends to make þese sorts of tasks, which depend on exactly þe sort of supervisory observation process Wayland restricts, harder to put togeþer, but clearly it's possible, or else Wayland wouldn't have screenshotting programs.

[-] eugenia@lemmy.ml -4 points 6 days ago

There's nothing like you ask. Most time tracker apps are just a calendar where you write manually how much time you spent on something. So you can use something like Kimai, or use a paper calendar and write on it.

But text log of the active window and a screencap, that's the stuff of Microsoft Recall AI nightmares that Linux developers wouldn't be keen to implement. What you're asking is intrusive AI for others. Maybe you need to actually learn to be punctual and write down your activities, or simply, buy a Snapdragon laptop with Windows AI on it. And even then, that info stays with the AI, I don't think it's shared much with the user.

[-] null_dot@lemmy.dbzer0.com 9 points 6 days ago

Most time tracker apps

That's not what I asked for.

use a paper calendar and write on it.

You don't really understand time tracking, I see.

But text log of the active window and a screencap, that's the stuff of Microsoft Recall AI nightmares

How is logging the title of the active window an AI nightmare ?

the stuff of Microsoft Recall AI nightmares that Linux developers wouldn't be keen to implement

Like this you mean? Yes, surely that doesn't exist.

Maybe you need to actually learn to be punctual and write down your activities

Maybe you need to try being... a bit less of a dick ?

buy a Snapdragon laptop with Windows AI on it

Kinda speechless at this one. Well done.

[+] BCsven@lemmy.ca -18 points 6 days ago* (last edited 6 days ago)

Chatgt says build a scrip using a few tools. xdotool and scrot. I don't know if this code is good or not. And some hashes are making markdown headers. How do we paste code on here?

#!/bin/bash

Set interval (in seconds)

INTERVAL=10

Output directories

LOG_FILE="$HOME/window_log.txt" IMG_DIR="$HOME/window_snaps" mkdir -p "$IMG_DIR"

while true; do # Get timestamp TS=$(date "+%Y-%m-%d_%H-%M-%S")

# Get active window title
TITLE=$(xdotool getactivewindow getwindowname 2>/dev/null)

# Fallback if title is empty
if [ -z "$TITLE" ]; then
    TITLE="(No active window)"
fi

# Take screenshot
IMG_FILE="$IMG_DIR/snap_$TS.png"
scrot "$IMG_FILE"

# Log entry
echo "$TS | $TITLE" >> "$LOG_FILE"

# Wait before next iteration
sleep $INTERVAL

done

[-] MasterOKhan@lemmy.ca 12 points 6 days ago

There is an etiquette to not just copy and pasting from ChatGPT. The fact you couldn’t verify the code yourself is a bigger issue.

I understand you may have thought this may help, it really does not.

[-] BCsven@lemmy.ca 0 points 6 days ago* (last edited 6 days ago)

Sure, seemed OP wasn't a google searcher so I tried to show an example of two programs coming together. The code seems plausible, but its there as an example of steps. I find LMM is typically trash, but can get you started. But your etiquette note is noted

[-] eshep@social.trom.tf 7 points 6 days ago

@BCsven
This is so much less helpful than just posting "I don't know" or "beats me".

First, if you're gonna post code, put it in a code block. And nevermind you not knowing if the ""code is good"", it doesn't even adhere to the question that was asked; the two programs you suggested are not even wayland compatible tools.
@null_dot

[-] BCsven@lemmy.ca 0 points 6 days ago

OP seemed like they didn't know where to start, so linking programs together was my suggestion. With a rough example. If thats against etiquette the noted.

As for helpfulness, where are the code block entries. I have preview, hyperlink, inage, bold, italic,quotes,lists and spoilers across the bottom, and no codeblock.

As for Wayland compatible this is where, somebody reads between the lines. If those two aren't Wayland compatible search for Wayland compatible tool like "xxxxxx".

[-] null_dot@lemmy.dbzer0.com 7 points 6 days ago

Maybe ask chatgpt whether xdotool is compatible with wayland.

I get that you're trying to help but, this is not the way.

[-] BCsven@lemmy.ca -1 points 6 days ago

i figured Id leave some searching to you, lol.

[-] null_dot@lemmy.dbzer0.com 2 points 6 days ago

This might shock you but... I have actually spent some time looking into this.

The tools you've suggested aren't compatible with Wayland. It seems that alternatives don't really exist, or cause the problems I mentioned in my post.

Additionally, I have a few decades experience with Linux and while I'm not some amazing Linux guru I do know what a bash script is and how to "link two programs together".

Finally, like everyone on the planet I also know what chatgpt is and might even consider using it to create a bash script if I knew what tools were compatible with Wayland.

[-] BCsven@lemmy.ca 0 points 6 days ago

No worries, the post made it sound like you had not tries searching. So I assumed your were a new person that had put in no effort yet. My mistake

[-] null_dot@lemmy.dbzer0.com 1 points 6 days ago

No worries, your comments make you sound insufferable.

[-] BCsven@lemmy.ca 0 points 5 days ago

Lol, this is why the Linux community gets a bad rap. Somebody tries to be helpful and the community gets hostile. Hope you have a good rest of your week dude.

[-] null_dot@lemmy.dbzer0.com 1 points 5 days ago

It's funny you say that. I was thinking exactly the same thing about your comments.

I've asked for help, you posted a chatgpt response, and now you've claimed eleventy times that I seem like I don't know where to start or don't seem like the type to search things.

"OP should've googled it first" is one of the hallmarks of toxic communities.

[-] monovergent@lemmy.ml 0 points 5 days ago

Pot, meet kettle

[-] RedWizard@hexbear.net -1 points 6 days ago

Look into EMACS and Org-Mode, it can help you do just that.

[-] null_dot@lemmy.dbzer0.com 1 points 6 days ago

emacs can periodally log the focussed window title ?

this post was submitted on 07 Oct 2025

62 points (100.0% liked)

Linux

57274 readers

735 users here now

From Wikipedia, the free encyclopedia

Linux is a family of open source Unix-like operating systems based on the Linux kernel, an operating system kernel first released on September 17, 1991 by Linus Torvalds. Linux is typically packaged in a Linux distribution (or distro for short).

Distributions include the Linux kernel and supporting system software and libraries, many of which are provided by the GNU Project. Many Linux distributions use the word "Linux" in their name, but the Free Software Foundation uses the name GNU/Linux to emphasize the importance of GNU software, causing some controversy.

Rules

Posts must be relevant to operating systems running the Linux kernel. GNU/Linux or otherwise.
No misinformation
No NSFW content
No hate speech, bigotry, etc

Related Communities

Community icon by Alpár-Etele Méder, licensed under CC BY 3.0

founded 6 years ago

MODERATORS

AgreeableLandscape@lemmy.ml

nooter692@lemmy.ml

MarcellusDrum@lemmy.ml

cypherpunks@lemmy.ml

cyclohexane@lemmy.ml

d3Xt3r@lemmy.nz