Last updated: 2026-10-21

📓 Daily Journal with Mood Tracking

A 3-minute evening check-in that captures wins, blockers, and mood — then surfaces trends weekly so you actually notice your patterns.

⏱ 45 minutes 💵 Model usage only 📊 easy ⭐ Hermes

The problem

You know you should journal but 'open a blank doc' is too much activation energy on day 63. Most journaling apps become guilt-inducing unused apps. The actual useful signal — are you happier or less happy this month than last — requires pattern detection across weeks, which nobody does manually.

The outcome

A short Telegram conversation each evening: 'What went well? What was hard? Mood 1–10?' Two minutes. The agent stores everything. Every Sunday it sends a trend summary: 'Mood is up 1.2 points vs last month. Workouts correlate with your best days. You've flagged the same blocker 4 weeks running — maybe time to escalate?'

Why Hermes

Hermes has persistent memory and cron jobs that deliver to Telegram (see our memory and scheduled-jobs guides), so a nightly prompt and a Sunday summary are two scheduled jobs and the agent can recall what you said months ago.

Alternatives worth considering

  • OpenClaw — If you'd rather the data stay local, have it append each entry to a markdown file on your machine instead of relying on cloud memory
  • Claude Cowork — A Claude Project named 'Journal' with the running log pasted in works for manual weekly check-ins

Setup steps

  1. Step 1: Define your check-in questions

    Keep it to 3–4 max. Standard set: wins, blockers, mood 1–10, one word for the day. You can customize — physical health, learning goals, relationships — but don't exceed 5 or you'll skip days.

  2. Step 2: Schedule the daily prompt

    Create a cron job that messages you at 9pm, for example: hermes cron create "0 21 * * *" "Ask me what went well today, what was hard, my mood 1-10 and one word for the day, then store the entry" --deliver telegram. Skipped days are fine; the agent should not nag.

  3. Step 3: Set up the Sunday trend review

    A second cron job each Sunday: pull the last 7 days, compute the mood average, highlight recurring themes and compare with the previous week and month. 200 words at most.

  4. Step 4: Add month-end deeper review (optional)

    On the 1st of each month, the agent writes a longer reflection: 'What were your 3 best days this month? What patterns do you see?' This is where real insight lives.

Example prompt

Ask me: what went well today, what was hard, my mood 1–10, and one word for the day. Store the entry. If it's Sunday, also generate a weekly trend summary comparing mood and themes to last week and last month.

Pitfalls to avoid

  • Over-engineering the prompts. 10 questions feels thorough but you'll quit by week 3. Ask fewer things and you'll have a year of data.
  • Not archiving entries. Memory systems can forget. Have the agent also write each entry to a simple markdown file in your Dropbox/iCloud as a backup.
  • Storing in a for-profit cloud without thought. This is sensitive data. Read the provider's data retention policy. If you're uncomfortable, use local storage + a local model.

Cost breakdown (monthly)

ItemCost
Hermes (open source, self-hosted)$0 for the software
Model calls (a short daily exchange plus a weekly summary)See the model pricing table

Cost: Model usage only. Model rates change often; check the model pricing table or the cost calculator for current figures.

Related guides

← Back to all use cases · Compare platforms at the decision guide.