oura-daily-watch

Generate daily Oura summaries and anomaly hints, optionally combined with Discord activity patterns.

Prereqs

• python3
• Env var:
  • OURA_PERSONAL_ACCESS_TOKEN

Fetch Oura metrics

Run:

python3 {baseDir}/scripts/oura_fetch.py --date today --tz Asia/Tokyo

For a specific day:

python3 {baseDir}/scripts/oura_fetch.py --date 2026-02-12 --tz Asia/Tokyo

The script returns JSON with:

• today: daily metrics snapshot (sleep/readiness/activity)
• baseline: 7-day baseline averages (where available)
• flags: anomaly flags (readiness drop, short sleep, RHR spike, etc.)

Combine with Discord behavior (optional but recommended)

When user wants mood/activation context, read channel messages and compute lightweight behavior signals:

• message count in last 24h
• late-night activity ratio (e.g. 00:00–05:00 JST)
• sudden spike vs recent average (best effort)

Use message.read for channel history, then summarize only aggregate stats (no sensitive content dump).

Output pattern

For morning post (short):

• Readiness / Sleep / Activity scores
• Sleep duration and any notable deviation from baseline
• 1-line behavior note from Discord activity (if collected)
• Advice:
  • normal day: short maintenance tip
  • anomaly day: concrete low-friction actions (rest, hydration, meeting load, nighttime cutoff)

Alerting policy

• Always include morning summary (daily)
• Add stronger advisory language when flags is non-empty
• If Oura fetch fails: output Oura未取得 + reason, and do not hallucinate values

Integration pattern: weather + Oura daily post

When the user wants Oura to appear in a daily weather post:

1. Keep weather as the first block.
2. Run oura_fetch.py and append a short Oura block.
3. If flags is empty: add 1-line maintenance advice.
4. If flags is non-empty: add concrete low-friction anomaly advice.
5. Keep total output concise (typically 8–14 lines).

Recommended heading:

• 【毎朝の天気+Oura】藤沢市・渋谷区

Safety

• Treat this as health-supporting insight, not medical diagnosis
• Avoid definitive medical claims
• Keep recommendations practical and reversible