Deployable templates for families, power users, and everyone in between. Pick a use case, copy the config, and go.
Set daily gaming caps with gentle warnings before enforcement kicks in.
Your child games too much and loses track of time. You want fair limits without constant arguments.
Leassh monitors running processes and vision-classified screenshots. When gaming time hits 80%, the child gets a warning notification. At 100%, the gaming app is closed automatically.
screen_time:
enabled: true
capture_interval_minutes: 5
time_limits:
- node: kids-pc
category: gaming
max_minutes_per_day: 120
enforcement:
- node: kids-pc
category: gaming
warn_percent: 80
action: kill_categoryGet a warm, factual summary of your child's computer habits every week.
You want to understand your child's computer habits without hovering over their shoulder or reading their messages.
Leassh captures periodic screenshots, classifies activities via AI vision, and generates a weekly narrative report. The tone is like a thoughtful teacher at a parent conference: positive, factual, and non-judgmental.
screen_time:
enabled: true
capture_interval_minutes: 10
data_retention_days: 30
vision:
ollama_url: http://localhost:11434
model: llava
# Call the insights API weekly:
# GET /fleet/api/insights/kids-pc/weeklyAutomatically log off the computer on school nights at a set time.
They stay up late on the computer, especially on school nights. Morning battles follow.
A time-based rule detects when someone is logged in past bedtime on weeknights and issues a warning, then forces a logoff after a grace period.
rules:
- name: bedtime_enforcement
when: "time >= 22:00 AND time < 06:00 AND day_of_week IN [Mon,Tue,Wed,Thu,Sun]"
nodes: [kids-pc]
actions:
- notify: warning
- ssh_command: "shutdown /l /f" # Windows logoff
notifications:
telegram:
bot_token: "your-bot-token"
chat_id: "your-chat-id"Reclaim your GPUs from family gaming for Ollama inference workloads.
Your family uses the GPUs you need for Ollama, ComfyUI, or training runs. You need shared machines to prioritize inference during work hours.
Leassh monitors GPU utilization across machines, blacklists games during work hours, and can start/stop Ollama services based on rules.
nodes:
- name: gpu-workstation-1
host: 192.168.1.50
ssh: user@192.168.1.50
gpu: nvidia
shared: true
blacklist:
global:
- steam.exe
- epicgameslauncher.exe
rules:
- name: work_hours_gpu_priority
when: "time >= 09:00 AND time < 18:00 AND day_of_week IN [Mon,Tue,Wed,Thu,Fri]"
actions:
- kill_blacklisted
- ssh_command: "systemctl start ollama"
- name: evening_free
when: "time >= 18:00"
actions:
- ssh_command: "systemctl stop ollama"Detect crashed services and restart them automatically with Telegram alerts.
Your home services crash and you don't notice for hours. Plex, Home Assistant, or Nextcloud go down silently.
Leassh probes your servers every few minutes, detects when critical processes disappear, restarts them via SSH, and sends you a notification.
nodes:
- name: homelab-server
host: 192.168.1.10
ssh: admin@192.168.1.10
rules:
- name: plex_recovery
when: "NOT process_running('plexmediaserver')"
nodes: [homelab-server]
actions:
- ssh_command: "systemctl restart plexmediaserver"
- notify: warning
- name: nextcloud_recovery
when: "NOT process_running('apache2')"
nodes: [homelab-server]
actions:
- ssh_command: "systemctl restart apache2"
- notify: warning
notifications:
telegram:
bot_token: "your-bot-token"
chat_id: "your-chat-id"Understand whether computer time is productive homework or passive entertainment.
They say they're doing homework, but are they really? You want data, not surveillance.
Deep vision analysis classifies not just the app but the content quality and engagement type. You see a breakdown of "active learning" vs "passive consumption" without reading their screen.
screen_time:
enabled: true
capture_interval_minutes: 5
vision:
ollama_url: http://localhost:11434
model: llava
# Use the deep analysis endpoint for richer data:
# GET /fleet/api/insights/kids-pc/deep
#
# Returns: content_quality, engagement_type,
# specific_content, new_or_notableKnow which platforms they use and how much, without reading their messages.
You want to know what social platforms your child uses without invading their privacy or reading their conversations.
Vision analysis identifies social media platforms from screenshots without capturing message content. You get time-per-platform data, not conversation logs.
screen_time:
enabled: true
capture_interval_minutes: 10
vision:
ollama_url: http://localhost:11434
model: llava
# The weekly report breaks down social media by platform.
# Deep analysis captures platform names in specific_content.
#
# Optional: set time limits on social media
time_limits:
- node: kids-pc
category: social_media
max_minutes_per_day: 60Auto-kill heavy apps left running on shared workstations after idle timeout.
People leave heavy apps running on shared workstations. Chrome with 200 tabs, Blender renders, or Docker containers eating all the RAM.
Leassh detects idle sessions and kills specified resource-heavy processes after a configurable timeout. Shared machine etiquette, enforced.
nodes:
- name: shared-workstation
host: 192.168.1.30
ssh: admin@192.168.1.30
shared: true
rules:
- name: idle_cleanup
when: "idle_minutes > 30"
nodes: [shared-workstation]
actions:
- ssh_command: "pkill -f 'blender|chrome|firefox'"
- notify: info
probe:
idle_threshold: 600 # 10 min for idle detectionMonitor that an elderly parent is using their computer normally, with alerts on pattern changes.
Is your parent using their computer normally? A sudden change in patterns could mean they need help.
Leassh establishes usage baselines over time. When patterns change significantly (no login by usual time, unusually long sessions, sudden drop in activity), you get alerted.
nodes:
- name: moms-pc
host: 192.168.1.20
ssh: user@192.168.1.20
screen_time:
enabled: true
capture_interval_minutes: 15
data_retention_days: 60
rules:
- name: no_morning_login
when: "time >= 11:00 AND NOT logged_in"
nodes: [moms-pc]
actions:
- notify: warning
notifications:
telegram:
bot_token: "your-bot-token"
chat_id: "your-chat-id"Self-monitoring mode: see where your time actually goes with AI classification.
Where does your time actually go? You think you spent 4 hours editing, but vision says 2 hours editing and 2 hours on Reddit.
Run leassh on your own machine in self-monitoring mode. No enforcement, no limits, just honest data about your computer usage patterns and a weekly summary.
nodes:
- name: my-workstation
host: localhost
ssh: me@localhost
permissions:
screenshot: true
screen_time:
enabled: true
capture_interval_minutes: 10
vision:
ollama_url: http://localhost:11434
model: llava
# No enforcement, no time limits.
# Just data and weekly reports.
# GET /fleet/api/insights/my-workstation/weekly