Mental Cares
Capture your moments, cultivate insights.
A journaling app for mental health. Write the day down with a track behind it, get a response from an AI that read what you actually wrote, watch your mood take shape as a curve — and, when you want company, hand it to an anonymous community that catches it.
Save
Why we built it
— Xu Pei-fen, I Only Worry Whether the Rain Will Keep Falling Until Tomorrow Morning. The line the whole project started from.
Some words come out. There is still nowhere for the feeling behind them to go.
So you wrap them carefully and post them to Threads or X. Nobody catches them — you get negative replies instead. If you are unlucky, your identity gets passed around and put on trial.
And the algorithms underneath, tuned for ad revenue rather than for you, quietly amplify the negativity, the echo chamber and the hostility.
So we built somewhere else to put it.
How it works
Pick a track from the in-app library that matches the day, write as much or as little as you want, and set where your mood actually sits on a scale from −1 to +1.
GPT-4, grounded in the clinical material the team compiled, replies to what you actually wrote — supportively — and notes how your mental state reads from it.
The entry joins your timeline and your emotional curve. Share it anonymously and the community can reply too, so the next reader is a person rather than a model.
The app
Feature 01
The composer is three controls and nothing else: a song, a text area that only asks what happens today…, and a mood slider. Hit Save and the entry goes off for analysis; the AI's reply comes back before you leave the screen.
Feature 02
Every entry lands on a vertical timeline, newest first, each one still carrying the song you chose that day. Open one and you get the whole record back: what you wrote, what the AI said, how it read your state, and anything the community replied.
Feature 03
The Statistics tab plots your Mood Index over time twice: the mood you reported when you wrote, and the sentiment the AI independently read out of the writing itself. The days where the two lines pull apart are the interesting ones.
Feature 04
Explore is a deck of other people's entries, swiped one at a time, with the writer's song still attached. No handles, no follower counts, no engagement metrics to farm — just an entry, the AI's reply, and whatever strangers wanted to say back.
Feature 05
Six conditions — depression, bipolar disorder, anxiety, PTSD, delusional disorder and schizophrenia — each with Symptom, Explanation and Commonly Affected Groups. The team compiled every article by hand from authoritative hospital sources rather than pulling it from Wikipedia or letting a model write it.
Built with
The client lives in this repository; the API, the LangChain retrieval layer and the sentiment models live in CCdiary-backend.
The award
1st
Prize
SITCON Hackathon 2024, on the Mental Care & Support track (心靈照護與關懷). Built by a team of four under the name 不嘻嘻的時候，黑客松松澤, shipping under vaclis.
The team
What's next
Surface entries from steadier community members to users whose curve has been trending negative for a while — so the people most likely to be dropped get caught first.
Partner with licensed counsellors and put a direct channel in the app, for the moment when a journal and a stranger's kindness are not the right tools.
Keep extending the emotional analytics so someone can see their own mental and physical patterns from the outside, rather than only from inside a bad week.
ccDiary is not publicly released. Two things are in the way: every entry costs a GPT-4 API call, and the backend needs hosting we can sustain. The build in this repository is the hackathon build, and the demo backend is currently offline.
ccDiary is also not a medical device, and nothing in it is a diagnosis. The app says so itself at the foot of every article it ships: “If the symptoms mentioned above persist and are pronounced, you should seek help from a specialist physician.” If you are in crisis in Canada or the US, call or text 9-8-8; in Taiwan, call 1925.
Original materials
Depression
Bipolar
Anxiety
Post-traumatic stress disorder
Delusional
Schizophrenia