Skill Catalog

Eight skills. One pipeline. Any MCP-compatible agent.

Soma skills are standard MCP tool calls that read your iPhone’s Apple Health and location data from a local server on your Mac. Use the pre-built skills below, or write your own.

FreeMulti-source intelligence

Concierge

Context-aware restaurant, coffee, and activity recommendations based on exact location and fatigue levels.

Concierge combines live GPS, your recent activity level, and learned time-of-day patterns to suggest places you'll actually want to go. No more dictating your location and mood — the agent already has the signal.

Data used

  • Current GPS location
  • Step count (last 24h)
  • Active energy (recent)
  • Heart rate (resting)
  • Time of day

Example interaction

You
Find me dinner.
Agent
You're on Fayetteville St, you've walked 14k steps, and you usually go for Asian on Thursdays. City Market Sushi is a 3-minute walk, no wait right now, fits your usual price range. Directions sent to your device.
Included with SomaGeofence trigger

Welcome Home

A briefing when you walk through the door. Time-aware, personalized, and context-rich.

Welcome Home fires on a geofence event at your home coordinates. The agent pulls your day's activity, what's coming up on your calendar, and unresolved items — and delivers a briefing tuned to the time of day.

Data used

  • Geofence events (home)
  • Step count (today)
  • Workouts (today)
  • Sleep score (last night)
  • Time of day

Example interaction

Agent
Welcome home. 12,400 steps, two workouts, HRV back above baseline. Dinner reservation at 7:30. You haven't started the return you meant to ship — it's still in the hallway.
Included with SomaHealthKit integration

Sleep-Adaptive Briefing

Morning brief that modulates detail length and tone based on last night's sleep quality score and stages.

Slept badly? Get a shorter, gentler brief with only the essentials. Slept great? Get the full rundown. Sleep-Adaptive Briefing reads last night's HealthKit sleep stages and score, then shapes the output accordingly.

Data used

  • Sleep score
  • Sleep stages (REM, core, deep)
  • Total sleep duration
  • Bedtime / wake time

Example interaction

You
Morning brief.
Agent
Rough night (sleep score 62, light on deep). Keeping this short: two meetings before noon, one blocker on the deploy, weather clear. Take it slow.
Included with SomaContinuous location

Proximity Nudges

Reminders for errands and places you care about, triggered by where you actually are.

Tell the agent what you need to do near certain places. Proximity Nudges fires only when you're actually there — not when you're doom-scrolling at your desk thinking about it.

Data used

  • Current GPS coordinates
  • Location history array
  • Geofence events
  • User-defined points of interest

Example interaction

Agent
You're passing a UPS Store — your Amazon return has been sitting in your bag since Tuesday.
Included with SomaBiometric analysis

Recovery Coach

Post-workout hydration, nutrition, and rest recommendations based on biometric recovery trajectory and HRV.

Recovery Coach pulls HRV trend, resting heart rate, and workout intensity to recommend how hard to push today. It catches the days your body is flagging before you feel it.

Data used

  • HRV (7-day trend)
  • Resting heart rate
  • Most recent workout intensity
  • Sleep score (last night)

Example interaction

You
How should I train today?
Agent
HRV is down 15% this week and you slept 5h 40m last night. Swap the strength session for mobility or a walk. Push tomorrow.
Included with SomaActivity state sync

Auto-Status

Automatic Slack or Discord status updates reflecting your real physical state (walking, driving, working out).

Auto-Status watches the Core Motion activity type and updates your Slack or Discord presence so your team knows when you're actually away from your desk versus deep-focused.

Data used

  • Activity type (walking / running / driving / stationary)
  • Workout in progress
  • Current location (coarse)

Example interaction

Agent
Status: 🏃 On a run — back in 30. (auto-set based on Core Motion workout state)
Included with SomaHistorical analysis

Trend Correlations

Pattern detection across your health data over time. Surfaces what's actually moving the needle.

Trend Correlations runs statistical analysis over your historical HealthKit data to surface non-obvious relationships — “alcohol on Friday correlates with HRV drops through Tuesday,” “leg days push your deep sleep up 18%.”

Data used

  • HRV history (90d)
  • Sleep stages history (90d)
  • Workout history (90d)
  • Step count history (90d)
  • Resting heart rate history (90d)

Example interaction

You
What's actually impacting my sleep?
Agent
Two strongest signals in the last 90 days: alcohol nights cut your deep sleep by 34% on average, and days with over 12,000 steps push your REM up by 22 minutes.
Included with SomaSpatial memory

Location Journal

Automatic time-tracking and location logging. No manual check-ins, no apps to open.

Location Journal quietly builds a searchable history of where you've been, how long you stayed, and how you got there. Query it in natural language: “when was the last time I was at the Raleigh office,” “how much time did I spend at the gym this month.”

Data used

  • Location history array
  • Geofence events
  • Activity type transitions
  • Timestamps

Example interaction

You
How much time did I spend at the gym this month?
Agent
14 hours 22 minutes across 19 visits. Average visit length: 45 minutes. Most consistent: Tuesday and Thursday mornings.

Build your own.

Every skill is a standard MCP tool call. If you can write a function, you can write a Soma skill.

Read the docs →