OpenCal

OpenCal is a calorie & nutrition tracker with a beautiful iOS app. This skill lets your AI agent log meals, check progress, and update goals — so the user never has to open the app to track what they eat.

“I just had a burrito for lunch” → agent searches, scales, logs it → it shows up in the app instantly.

Setup

1. Download OpenCal from the App Store
2. Sign in and set your calorie/macro goals
3. Go to Profile → API Keys → Generate
4. Save the key permanently (add to your shell profile so it persists across sessions):

   echo 'export OPENCAL_API_KEY="sk_your-key-here"' >> ~/.zshrc
   source ~/.zshrc
   

   On bash, use ~/.bashrc instead. The key must survive terminal restarts — a bare export only lasts one session.

When to use this skill

Log a meal

The most common flow. User says something like “I had 200g chicken breast for lunch”:

1. Search for the food:

curl -s "${OPENCAL_BASE_URL:-https://api.opencal.ai}/api/v1/food/search?q=chicken+breast&limit=5" \
  -H "Authorization: Bearer $OPENCAL_API_KEY" | jq '.results[] | {name, calories, protein, carbs, fat}'

Results are per 100g. Pick the best match.

2. Scale to the actual amount and log:

curl -s -X POST "${OPENCAL_BASE_URL:-https://api.opencal.ai}/api/v1/food/log" \
  -H "Authorization: Bearer $OPENCAL_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "Chicken Breast",
    "amount": 200, "unit": "g",
    "calories": 330, "protein": 62, "carbs": 0, "fat": 7.2,
    "mealType": "lunch"
  }'

Multiply all nutrition values by amount / 100. The entry appears in the app immediately.

Use mealType: breakfast, lunch, dinner, or snack. Add "loggedAt": "2026-02-18T12:00:00Z" to backfill a past meal (defaults to now).

3. Confirm with the user: “Logged 200g chicken breast for lunch — 330 kcal, 62g protein.”

Check daily progress

curl -s "${OPENCAL_BASE_URL:-https://api.opencal.ai}/api/v1/food/log?date=2026-02-18" \
  -H "Authorization: Bearer $OPENCAL_API_KEY" | jq '{totals, entries: [.entries[] | {name, calories, mealType}]}'

Omit ?date= for today. Returns each entry plus totals (calories, protein, carbs, fat).

Check and update goals

# Current goals
curl -s "${OPENCAL_BASE_URL:-https://api.opencal.ai}/api/v1/goals" \
  -H "Authorization: Bearer $OPENCAL_API_KEY" | jq

# Update (only include fields to change)
curl -s -X PUT "${OPENCAL_BASE_URL:-https://api.opencal.ai}/api/v1/goals" \
  -H "Authorization: Bearer $OPENCAL_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"calorieGoal": 1800, "proteinGrams": 160}'

Delete a log entry

Get the entry ID from the daily log, then:

curl -s -X DELETE "${OPENCAL_BASE_URL:-https://api.opencal.ai}/api/v1/food/log/{id}" \
  -H "Authorization: Bearer $OPENCAL_API_KEY"

Smart search — finding the right product

The database is primarily English (USDA + generic foods). Follow this strategy:

1. Search the product name as-is first. If the user says “Rinderhackfleisch mager”, search that.
2. If zero results, translate to English and retry. “Rinderhackfleisch mager” → search “ground beef lean”. This covers most cases.
3. If multiple results, pick the closest match. Compare fat %, brand, or description. Go with the best fit.
4. If confidence is low, tell the user. Example: “I found 3 options for ground beef — went with 90% lean (170 kcal/100g). Here are the others if that’s not right: [list]. You can also correct it in the app.”
5. Never hardcode nutrition values from memory. Always search first. The database has scaled, verified values.
6. Try simpler/shorter terms if specific ones fail. “Rewe Kultur Heidelbeeren” → “blueberries”. “More Nutrition Protein Sahne” → “whey protein powder”.

The goal: use real DB entries whenever possible. The user can correct in the app if the match isn’t perfect — that’s better than made-up numbers.

Important notes

• Search results are per 100g — always scale before logging
• If the user doesn’t mention an amount, ask — don’t guess
• Always confirm what you logged so the user can correct mistakes
• Rate limit: 100 requests/min

Updating this skill

npx skills update

This pulls the latest version from the OpenCal skill repo. Run npx skills check to see if an update is available.