Drive Efficiency Coach

When to use

Activate this skill when the user asks anything along the lines of:

• "Am I driving efficiently?"
• "Why does my range vary so much?"
• "How can I improve my Wh/km?"
• "Was my last trip efficient?"
• "Show me my efficiency over the last month."

If the user only asks for a charge cost summary or battery health, prefer the charge_review or battery_health_report MCP prompts instead.

Prerequisites

You need a TeslaMate car_id. If unknown:

1. Call teslamate_get_cars to list available cars.
2. If exactly one is returned, use it silently.
3. If multiple, ask the user which one before continuing.

Workflow

1. Pull the dataset

Call teslamate_get_car_drives with:

• car_id: target car
• start_date: 30 days ago in RFC3339 (e.g. 2026-04-06T00:00:00Z)
• end_date: now in RFC3339
• min_distance: 5 (filters out parking-lot moves where Wh/km is meaningless)
• fetch_all: True whenever the window is longer than ~1 month — the

TeslaMate API returns at most 100 entries per page, and the wrapper will auto-paginate to give you the complete set.

If the user specifies a different period (e.g. "this year", "last week"), use that instead. Always pass both start_date and end_date. For a year-long window, fetch_all=True is mandatory or you'll undercount.

Also pull teslamate_get_car_charges over the same window — needed for the timeline (Charge events) and to score the Recharge dimension.

2. Compute per-drive efficiency

For each drive in the response:

• Distance: prefer the field expressed in km. If TeslaMate is configured

for miles, convert (1 mi = 1.609 km).

• Energy used: derive from whichever field is present —

typically consumption_kwh, or start_ideal_range_km - end_ideal_range_km multiplied by the car's nominal Wh/km, or start_battery_level - end_battery_level multiplied by usable pack capacity. Use whichever yields the cleanest number; document the source in the answer.

• wh_per_km = energy_used_wh / distance_km
• avg_speed_kmh = distance_km / (duration_min / 60)
• Skip drives where distance_km < 5 or energy_used_wh <= 0.

3. Establish a baseline

From the cleaned set of drives:

• median Wh/km → user's normal
• p75 Wh/km → "above normal"
• p90 Wh/km → "outlier"
• best drive (lowest Wh/km on a meaningful distance, ≥ 20 km)
• worst drive (highest Wh/km on ≥ 5 km)

4. Identify outliers

Flag drives with wh_per_km > p90. For each, surface:

• date and time
• distance
• duration (min)
• average speed (distance / duration)
• Wh/km
• outside temperature if exposed
• start/end SoC if useful

5. Cross-reference patterns

Group outliers by:

• Speed band derived from average speed:
• city: < 50 km/h
• mixed: 50–90 km/h
• highway: > 90 km/h
• Temperature (if available): cold (< 5°C), mild (5–25°C), hot (> 25°C)
• Trip length: short (< 15 km), medium (15–60), long (> 60)

6. Coach

Pick the 1–3 most actionable insights, written as direct advice:

• "Highway drives above 110 km/h average ~X% more Wh/km than your baseline.

Cruising at 100–110 saves a meaningful chunk."

• "Cold mornings push your consumption up by ~X%. Precondition while

plugged in to recover most of it."

• "Short trips under 10 km show high Wh/km because cabin warm-up isn't

amortized — combine errands when you can."

Search true insights from data. Avoid generic platitudes. Anchor each insight in a concrete number derived from the user's own data.

7. Render (always)

Always finish with a visual report:

1. Call generative_search_prefab_components to discover what's

available now.

1. Call generative_generate_prefab_ui with a Dashboard layout

(outermost). Use the discovered components — favour MORE rather than fewer for an information-dense coaching dashboard.

1. If you want to map the outlier drives' start/end, call show_map

with their start_address / end_address — not a Prefab Map component (none exists).

Pitfalls

• Short drives mislead. Drives < 5 km are dominated by climate/start-up

energy, not driving style. Always filter them out before computing baselines.

• Energy data is sometimes missing. TeslaMate has gaps. Skip drives

with missing energy rather than imputing zeros.

• Units vary per TeslaMate install. Default to km, but if distances

look suspicious (e.g. all values < 5 for normal commutes), inspect a sample drive's units field and convert.

• Don't compare across very different periods. A "winter month vs

summer month" comparison needs explicit framing — temperature alone swings Wh/km by 20–30%.