create_habit
Play demoTry asking
- Create a habit called Meditate
- add a new habit: drink water
- track a habit called Read
- start tracking Exercise
ghost skill run ghost/habit-tracker create_habit Loop Habit Tracker (F-Droid): create a yes/no habit, tick or untick a day, list habits and read a habit's score and total count. Offline, no account.
Agent run: stopped early without the skill, finished with it (not independently checked).
One recorded run each way, claude-haiku-5-5, same phone and same prompt. Watch it in the Before / after tab.
41× fewer tokens, 1.6× faster with the skill. Agent run: stopped early without the skill, finished with it (not independently checked).
| Run | Time | Tokens | Tool calls | Result |
|---|---|---|---|---|
| No skill: agent drives by hand | 58.1 s | 1,562,778 | 31 | stopped early |
| With skill: find_skill, run_skill | 35.4 s | 38,069 | 5 | finished |
Prompt: Create a habit called Meditate, mark it done today and show me its stats.
Same prompt, same phone, clean app state per side. Left: no skill, the agent drives the UI by hand. Right: find_skill + run_skill. 'ok' means the agent finished its turn; the result was not independently checked.
One recorded pair on samsung-a03s, claude-haiku-5-5; tokens count every input, cache and output token once per assistant message. Not a signed run: it adds no run evidence.
Control Loop Habit Tracker on Android with an AI agent. Create habits, tick days and read your score and totals on your phone by asking an agent.
Target search phrase: AI agent habit tracker Android
Demo shows: create_habit (name=Meditate)
Ghost in the Droid seed skill for Loop Habit Tracker (
org.isoron.uhabits), the popular open-source habit tracker with scores and charts from F-Droid. Offline, no account.
| Action | What it does | Params |
|---|---|---|
create_habit | Create a yes/no habit | name, question |
check_habit | Mark a day done or missed (day 0 = today ... 3) | name, day, result |
get_habits | List the habit names | none |
get_habit_stats | Read score, month/year change, total count, frequency | name |
An agent can drive the whole app with these; compose them.
| Role | Action |
|---|---|
| Entry | create_habit |
| Primitives | check_habit |
| Read state | get_habits, get_habit_stats |
The app: F-Droid, or https://f-droid.org/packages/org.isoron.uhabits/. The skill:
pip install "ghost-in-the-droid>=1.6.0" # 1.6.0 is not on PyPI yet: until it ships, git clone the repo and `pip install -e .`
ghost skill install ghost/habit-tracker
ghost skill run ghost/habit-tracker create_habit --param name=Meditate --device <serial>
ghost skill run ghost/habit-tracker check_habit --param name=Meditate --param result=done --device <serial>
run_skill(name="ghost/habit-tracker", action="get_habit_stats", params={"name": "Meditate"})
# {"stats": {"score": "5%", "total": "1", "month": "+5%", "year": "+5%", "frequency": "Every day"}}
Try asking: "Create a habit called Meditate, mark it done today and show me its stats."
Real Samsung phone, Android 13 (720x1600, 300 dpi, en-GB), Loop Habit Tracker 2.3.2. License: MIT.
| App version | Android | Device | Skill version | Dates |
|---|---|---|---|---|
| 2.3.2 | 13 | Samsung real phone (720x1600, 300 dpi, en-GB) | 1.0.0 | 2026-11-02 |
yesBtn/noBtn ids.yes/no into booleans, so the result enum is done/missed.skip) is handled in the open_list subroutine; the notification permission is pre-granted.SOFT.md is the narrative an agent reads when the exact steps no longer match your app version and it has to adapt the skill.
What each action means, which UI landmarks it relies on and how success is recognised, so an agent can adapt it to another app of this kind or another version.
Package: org.isoron.uhabits (Loop Habit Tracker, GPL-3.0, version 2.3.2 from the official F-Droid repository, sha256-verified)
Verified: 2026-11-02 on a real Samsung phone, Android 13, 720x1600 @ 300 dpi, en-GB. Offline, no account.
Skill class: HARD. No login, no checkpoints.
skip). List screen: actionCreateHabit (+), action_filter; one row per habit (the name is a TextView) followed by four day cells (today, yesterday, ...) at x about 400/490/580/670 on a 720 px screen.buttonYesNo ("Yes or No") > nameInput, optional questionInput > buttonSave.yesBtn (done) and noBtn (missed); the check marks themselves are drawn, not text.scoreLabel (e.g. 5%), monthDiffLabel, yearDiffLabel, totalCountLabel, frequencyLabel, charts.create_habit(name, question)Yes or No > type the name (and question) > SAVE. Success: the name shows in the list.
check_habit(name, day, result)Tap the habit's day cell, wait for the dialog, tap yesBtn (done) or noBtn (missed). Result is done or missed (yes/no and true/false are accepted; the CLI turns bare yes/no into booleans).
get_habits()Reads the name TextViews left of the day cells.
get_habit_stats(name)Open the habit, read the overview labels, go back. Returns stats.
Other habit trackers also split a list and a detail screen; the day-cell positions and the done/missed dialog are the app-specific parts.
Try a sample prompt or run an action by name. Replace any parameters in braces with your own values.
create_habitTry asking
ghost skill run ghost/habit-tracker create_habit check_habitTry asking
ghost skill run ghost/habit-tracker check_habit get_habitsTry asking
ghost skill run ghost/habit-tracker get_habits get_habit_statsTry asking
ghost skill run ghost/habit-tracker get_habit_stats Each release is an immutable git tag. The tree hash pins the exact skill directory the installer verifies.
| Version | Released | App versions | Tree | Size | Provenance |
|---|---|---|---|---|---|
1.0.0latest | never | >=2.3.2 | pending | 16.7 KB | untagged |