Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Weekly or Monthly Health Report from Your Garmin Data

    2

    **What it does** Ask how your health has been lately, and the agent first checks which Garmin data tools are connected and reads their schemas.

    $5

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    Weekly or Monthly Health Report from Your Garmin Data

    Weekly or Monthly Health Report from Your Garmin Data

    Example session with this skill installed

    How's my health been this past week? Be straight with me about what's off.

    • Read your context and instructions
    • Compiled the weekly or monthly

    🏥 Health Report — Weekly (Wed Oct 14 – Tue Oct 20, 2026)

    Overall Health Trend: 🟢 77/100 — heart and Body Battery solid, HRV recovering but just under your normal range, sleep short, recorded steps very low

    DomainScoreStatusWeight this week
    😴 Sleep71🟡 Attention33%
    🧠 Stress & Recovery86🟢 Healthy28%
    ❤️ Heart Health89🟢 Healthy22%
    🚶 Activity60🟡 Attention17%
    ⚖️ Body Composition—no data0% (spread across the others)
    Overall77🟢

    This rests on the 4 best-recorded days (Wed, Thu, Fri, Tue), each with a recorded sleep and 246–295 stress readings. The watch missed 3 of 7 nights: Sat and Mon were only partly recorded (10 and 41 stress readings), and Sun has no wrist data (only 7.6 km and 191 active calories logged). Even on the best-recorded days, heart-rate data cover only about 14–19 of the 24 hours, so part of each waking day is missing. Daily averages use the 4 best-recorded days. Weekly totals (intensity minutes, active days) use all 7. I applied the adult 18–30 bands (age from your Garmin profile). The scores are a wellness heuristic for comparing you with yourself over time, not a clinical grade. Times are your watch's local time.

    😴 Sleep — 🟡

    Duration 52 · Consistency 81 · Quality 84 → 71

    DateAsleepFell asleepWokeSleep scoreDeep / Light / REM
    Wed Oct 147 h 14 m6:47 AM2:07 PM8724 / 56 / 20 %
    Thu Oct 155 h 20 m7:05 AM12:25 PM6440 / 56 / 4 %
    Fri Oct 166 h 34 m6:53 AM1:36 PM8237 / 47 / 16 %
    Sat Oct 17————not recorded
    Sun Oct 18————not recorded
    Mon Oct 19————not recorded
    Tue Oct 205 h 14 m6:34 AM11:49 AM7728 / 47 / 26 %

    On the 4 nights the watch recorded, you averaged

    6 h 06 min asleep. That's under the 7-hour adult guideline, and two of the nights were about 5¼ h. Against a 7 h target you're

    3 h 52 min short across those four nights (7 h 38 min short against 8 h). Trend: ⬇️ −11% (last three recorded nights vs first three). Four nights is thin evidence, though.

    Your timing is late but very steady. You fell asleep between 6:34 and 7:05 AM every recorded night (average 6:50 AM, spread ±13 min), and that regularity counts for more than the clock time. Wake time is looser: anywhere from 11:49 AM to 2:07 PM (±64 min). Because bedtime barely moves, the short nights were simply the early wake-ups. The Consistency score of 81 averages bedtime regularity (96) and wake-time regularity (67); on bedtime alone, Sleep would score 75 instead of 71.

    Quality is fine. Garmin's sleep score averaged 77.5 (range 64–87), deep sleep averaged 32%, and you were awake less than 10 minutes every night. REM averaged 17%, a bit under the typical 20–25%, mostly because of Thursday's 4%. Wrist sleep staging is rough, so don't read much into one night.

    What the data shows: Your sleep averages 6 h 06 min because you wake, on average, only about 6 h after a very steady ~6:50 AM lights-out, and two nights ended before 12:30 PM.

    Why it matters: Adults need 7+ hours regularly, and repeated short nights carry health and performance costs, including the recovery your training relies on.

    What to try: Give yourself a fixed 7 h 30 min sleep window. Keep your usual lights-out (~6:50 AM) and set one alarm for 2:20 PM every day. If your day has to start earlier, move lights-out 15 minutes earlier every three nights until the window fits.

    🧠 Stress & Recovery — 🟢

    Stress level 85 · Body Battery 96 · HRV status 72 → 86

    DateAvg stressRest %Body Battery highCharged / drainedOvernight HRVHRV status
    Wed Oct 143261%76+59 / −4140 msLow
    Thu Oct 152666%100+16 / −3647 msLow
    Fri Oct 162664%93+63 / −6051 msLow
    Sat Oct 17(39, only 10 readings)—(no overnight data)+0 / −5—Low
    Sun Oct 18no data—————
    Mon Oct 19(54, only 41 readings)—(no overnight data)+0 / −18—Unbalanced
    Tue Oct 202760%95+41 / −4149 msUnbalanced

    Stress: You averaged

    28 on the best-recorded days, inside Garmin's low band. Of the measured time, about 63% was at rest, 24% low, 11% medium and under 2% high. The trend is ↘️ −6%, which is the good direction. Garmin's own weekly figure is 34, which equals the average of all six days with any readings, including the sparse Sat (10 readings) and Mon (41 readings).

    Body Battery: It peaked at 76–100 after the four recorded nights (average 91), and sleep refilled about 79% of the available headroom (Wed 60%, Fri 89%, Tue 88%). From that high it drained 23–56 points by the day's last reading (average 40), and each day's readings end hours before your lights-out. Thursday's 100 doesn't fit a day that only charged +16, so treat it loosely. Without it the average high is 88, which falls in the same band.

    HRV is the soft spot. Garmin rated it Low Wed–Sat and Unbalanced Mon–Tue. Your 7-day average rose from 40 to

    47 ms, but that's still just under your personal balanced range of 48–67 ms. You're recovering, not fully recovered.

    Garmin's training readiness fell to 1 on Tuesday after you woke, following Monday's hard session (64 h recovery time). That's a training-load question more than a health one, and a training-analysis skill is the right place to plan around it. If you've also felt run down or unwell this week, mention the run of Low HRV readings to a doctor.

    ❤️ Heart Health — 🟢

    Resting HR 97 · HRV level 73 · SpO2 91 → 89

    DateResting HRSpO2 asleep (lowest single reading)Breathing rate asleep
    Wed Oct 1443 bpm95% (86%)17 /min
    Thu Oct 1553 bpm97% (93%)16 /min
    Fri Oct 1653 bpm96% (87%)17 /min
    Sat Oct 1758 bpm (partial wear)——
    Sun Oct 18———
    Mon Oct 1957 bpm (partial wear)——
    Tue Oct 2052 bpm98% (92%)16 /min

    Resting HR averaged

    50 bpm on the best-recorded days, which is typical for someone who trains. Three of those days sat at 52–53. Wednesday's 43 (overnight minimum 32) is one unusually low day, not an outlier by this report's 2-SD rule, and one day isn't a trend. Garmin's rolling 7-day resting HR eased from 56 to 53 over the week. The trend formula shows ↗️ +3 bpm, but only because of that Wednesday low, and a +3.0 bpm change doesn't pass the scoring's 3 bpm threshold, so there's no score penalty and no concern flag.

    HRV level: a 7-day average of 47 ms is a reasonable absolute level. HRV is very individual, though, and the trend against your own baseline (see Stress & Recovery) matters more.

    SpO2 averaged

    96.5% during sleep across the 4 nights. Lowest single readings of 86–93% are common with wrist sensors.

    Coverage caveat: readings stopped at about 10 AM every day, which is when your saved sleep window (midnight–10 AM) ends. So only 44–65% of each night was measured, and the last 1 h 49 m to 4 h 07 m of each sleep had no readings at all. Readings also began up to ~7 h before you fell asleep. To fix this, set the sleep window in Garmin Connect's sleep-time setting to roughly 6:30 AM–2:30 PM so it matches when you actually sleep.

    • Breathing rate asleep was steady at 16–17 breaths/min.

    🚶 Activity — 🟡

    Steps 15 · Intensity minutes 100 · Consistency 69 → 60

    DateStepsModerate minVigorous min
    Wed Oct 141,0364240
    Thu Oct 153,06600
    Fri Oct 162,535282
    Sat Oct 171,128 (partial wear)00
    Sun Oct 180 (no wrist data)00
    Mon Oct 19682 (partial wear)1172
    Tue Oct 2096500
    Week9,41281114
    7 days endingAvg steps/dayGarmin weekly stress
    Sep 293,15336
    Oct 63,25229
    Oct 135,62341
    Oct 20 (this week)1,34534

    Intensity minutes: 309 for the week, counted as moderate + 2 × vigorous (81 + 2 × 114). That's above the WHO range of 150–300. The unweighted sum (moderate + vigorous, the way this server totals its weeks) would be 195. Your last four full calendar weeks (Sep 21 – Oct 18) came to 317–589 on the same basis. Exercise is not your problem.

    Consistency: 3 active days (Wed, Fri, Mon), spread across the week but only 3 of 7. Sunday shows 7.6 km and 191 active calories logged with no wrist data, so any intensity minutes from that day aren't counted here.

    Steps: this is the weak spot. Steps don't capture non-walking exercise. On Wednesday and Monday you logged about 30 km of distance on only 1,036 and 682 steps. So this score reflects everyday walking only. Your recorded steps are very low, but part of that may be time the watch was off during the day, which the data can't separate from sitting still.

    What the data shows: Your recorded steps are very low, about 1,900/day on the best-recorded days (1,345 across all 7) against 3,153–5,623/day in the previous three weeks, though part of that may be daytime hours when the watch wasn't worn.

    Why it matters: Large cohort studies see much of the measured benefit of daily walking by around 7,000 steps/day, and because they count only steps, this score describes your everyday walking, not your overall activity, which is well above the WHO range.

    What to try: Take two 15-minute walks every day, one soon after waking and one after your main meal. Each adds roughly 1,500–2,000 steps, which would bring a typical day to about 5,000–6,000.

    ⚖️ Body Composition — no data

    You logged no weigh-ins this week, so this domain isn't scored and its 10% weight was spread across the other four.

    📋 Key Takeaways

    Get to 7+ hours. Your bedtime is already steady, so the fix is a fixed wake time that allows a 7 h 30 min sleep window (alarm ~2:20 PM with your current lights-out, or move lights-out earlier in 15-minute steps).
    2. Wear the watch every night and through the day, and match its sleep window to your real sleep (~6:30 AM–2:30 PM). Three of seven nights went unrecorded, SpO2 missed roughly the last 2–4 hours of each recorded night, and daytime gaps leave the step count uncertain. That's the biggest blind spot in this report.
    3.

    Walk twice a day for 15 minutes. Your training volume is high, but recorded everyday movement is very low (about 1,900 steps on the best-recorded days).

    Data not available this period

    • Sleep, overnight HRV and SpO2 for the Sat, Sun and Mon nights (Oct 17–19): no sleep was recorded. Sat and Mon had only 10 and 41 stress readings, and Sun had no wrist data at all.
    • Part of each waking day, even on the best-recorded days: heart-rate data cover about 14–19 of the 24 hours, so steps and stress from the unrecorded hours are missing, and the data can't tell time off-wrist from sitting still.
    • SpO2 covers only part of each recorded night (it stops at the 10 AM end of your saved sleep window).
    • Body composition: no weigh-ins logged.

    Wellness analysis of wearable data, not a medical assessment.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    What it does

    Ask how your health has been lately, and the agent first checks which Garmin data tools are connected and reads their schemas. It then collects one data point per day: all 7 days for a weekly report; for a monthly report, the last 7 days daily, four weeks of weekly step, stress and intensity aggregates, and every third day for the rest (or a compact range tool where one exists). From those values it works out the numbers itself: average sleep and sleep debt against a 7 to 8 hour target, bedtime and wake-time consistency in local time with the midnight wrap handled, sleep-stage share, average stress and the split of time between rest and low, medium and high stress, morning Body Battery, overnight recharge and daytime drain, the resting-HR trend over the period, HRV against your personal baseline (from Garmin's HRV status), multi-night SpO2 and whether the readings cover your actual sleep, steps, WHO-equivalent intensity minutes, active days, and weight trend if you log weight. Each of five domains (Sleep, Stress & Recovery, Heart Health, Activity, Body Composition) gets a 0-100 score and a green, yellow or red light. Every yellow or red domain gets three lines: what the data shows, why it matters, and one specific thing to try. The report ends with the top three actions, a "Worth discussing with a doctor" section when one of its flags applies (it names patterns and dates, never a diagnosis), a list of the data that was not available, and one disclaimer.

    Why this instead of a prompt

    A one-line prompt leaves every judgment to the agent: which days to pull, which Garmin totals to trust, what counts as good, and what to do when a night is missing. This skill writes those decisions down. The scoring bands live in a reference file that marks which anchors come from published guidance (AASM/SRS sleep duration, AHA and Mayo Clinic resting heart rate, WHO 2020 physical activity, two step-count meta-analyses, Garmin's own manuals) and which are heuristics. Days with no data are left out of the averages and listed as missing, a metric with no data for the whole period is skipped and its weight redistributed, and the skill never fills a gap with a typical value. It computes intensity minutes as moderate + 2 × vigorous instead of trusting a pre-computed total. In the demo that mattered: the connected server totals its weeks as moderate + vigorous without doubling, and that unweighted sum (195 minutes for the week) would have landed in a lower band than the correct 309. It also checks whether SpO2 readings cover your real sleep; in the demo they stopped at the end of the sleep window saved in the Garmin profile, so only 44 to 65% of each night was measured, and the report says how to fix the setting. It also holds a medical line: a written list of patterns that should go to a doctor, an override for urgent symptoms, and rules for when the adult benchmarks should not be applied at all. If you only want your Garmin numbers fetched and charted, a free skill may be enough. What this one adds is the written method: scores against a cited rubric, explicit handling of missing days, written doctor and urgent-symptom rules, and no handling of your Garmin credentials, because it reuses the Garmin MCP server you already have.

    Key features

    • Capability discovery: maps Garmin tools by function, using the open-source Taxuspt/garmin_mcp tool names as the reference. A function your server lacks is skipped, its weight redistributed, and the gap named in the report.
    • Two modes: weekly (the default, last 7 complete days) and monthly (30 days, with a sampling plan that keeps the context small). The mode is picked from your wording in English or Chinese (for example "lately", "over time", 最近, 这段时间).
    • Written rubrics for 14 scored components across five domains, weighted Sleep 30%, Stress & Recovery 25%, Heart Health 20%, Activity 15%, Body Composition 10%, with age adjustments for HRV, deep sleep and steps. The bands are calibrated for active adults aged 18 to 30, and resting HR has no age adjustment (see Known limitations).
    • Derived metrics computed by the agent rather than read off a dashboard: bedtime and wake-time standard deviation, sleep debt, Body Battery recovery efficiency, endpoint trends, 2-SD anomaly checks, and WHO-equivalent intensity minutes.
    • HRV read two ways: trend against your personal baseline (mapped from Garmin's HRV status) and the absolute 7-day level, with a rule never to call HRV low from the absolute number alone.
    • SpO2 coverage check: compares the hours that have SpO2 readings with when you actually slept, and if they don't line up, tells you to match the sleep window saved in your Garmin profile to your real sleep.
    • Doctor flags with specific triggers: overnight SpO2 persistently below 94%, resting HR regularly above 100 or up 15+ bpm for several days without an obvious cause, HRV Low or Poor while you feel unwell, an irregular-rhythm or abnormal ECG alert, trouble sleeping 3+ nights a week for 3+ months, unintentional weight loss over 5% in 6-12 months, and high stress together with low mood.
    • Urgent symptoms override the data: chest pain, fainting, severe shortness of breath, sudden weakness, numbness, confusion, trouble speaking or thoughts of self-harm stop the analysis and point to emergency services (and a crisis line such as 988 in the US).
    • Absolute benchmarks are switched off, and only trends described, for users under 18, pregnant users, people with a known heart, lung or sleep condition, and people on medication that changes heart rate.
    • Neutral body-composition comments and no weight-loss targets unless you ask for them.
    • Charts when your client can render them (a domain dashboard, sleep trend, stress and recovery; heart and activity trends added for monthly reports), compact tables otherwise.
    • Read-only: it never calls the write tools some Garmin servers expose (weight, blood pressure, food logging), and it does not write your health data to files or other services without asking.
    • Replies in the language you write in, English or Chinese.

    Who it is for

    Garmin wearers who want a periodic, plain-language read on their sleep, stress, recovery and heart data instead of a one-day dashboard. The score bands fit active adults aged 18 to 30 best; older or less active users still get the trends, but should read the resting-HR color with the calibration limitation below in mind. It suits people with non-standard schedules, since it judges sleep by duration and consistency rather than by what time you go to bed. It also looks at the health side separately from training for athletes (training load and performance are left to a training-analysis skill), with one caveat: the steps score doesn't credit sports that log few steps, such as cycling or swimming, so an athlete can get a yellow Activity score despite plenty of training, as in the demo.

    How it is built

    Two plain Markdown files: SKILL.md (343 lines: data sources, workflow, scoring weights, output template, medical boundaries) and references/health-benchmarks.md (363 lines: scoring rubrics, trend and anomaly methods, age adjustments, sources). No scripts, no dependencies, no bundled data, no API keys. It makes no network calls of its own; all data comes through the Garmin MCP server you have connected.

    Known limitations

    • It needs a connected Garmin data source, such as a Garmin MCP server (for example Taxuspt/garmin_mcp). Without one it produces no scores, charts or trends; it offers a conversation, or an analysis of numbers you paste, labeled as user-supplied.
    • Wrist sensors are estimates, and Body Battery and the stress score are Garmin's own proprietary measures. The scores compare you with yourself over time; they are not a clinical grade.
    • Score bands are calibrated for active adults aged 18 to 30. HRV and deep sleep are adjusted for age, and steps from age 60. Resting-HR bands are not, so a resting HR in the clinically normal 73 to 100 bpm range scores yellow (73 to 80) or red (above 80). If you are older or less active, read the resting-HR part of Heart Health mainly for its direction over time rather than its color.
    • There is no rule for days the watch was only partly worn, and no wear-coverage check before steps are scored. Garmin returns normal-looking values for days with only 10 to 41 stress readings, against 246 to 295 on the demo's best-recorded days. Even those best-recorded days had heart-rate data for only about 14 to 19 of 24 hours, so a low step count can partly reflect time the watch was off. In the demo, how the days were split and the wear caveat are judgment calls, not rules in the skill.
    • There is no rule for Garmin values that don't add up. In the demo, one day's Body Battery high of 100 didn't fit the +16 Garmin reported as charged that day; the report flagged it and gave the average with and without it, a judgment call the skill does not spell out.
    • The steps score does not account for sports that log no steps. In the demo an athlete with 309 WHO-equivalent intensity minutes still got a yellow Activity score (60), because the two high-distance days logged about 30 km each on fewer than 1,100 steps.
    • SKILL.md asks for the spread of both bedtime and wake time, but the rubric's consistency bands are written for bedtime only, so how the two are combined is left to the agent. The demo averaged them; bedtime alone would have scored Sleep 75 instead of 71, and the report says so.
    • Weekly reports have no previous-week baseline for sleep, resting HR or SpO2, and no minimum number of nights before a sleep score is marked low-confidence. The demo scored sleep on 4 of 7 nights and noted only that four nights is thin evidence.
    • The overall light has no rule beyond the per-domain bands, and the domain weights can bury a single finding. In the demo the overall came out at 77 (green) with two yellow domains, and six days of Low or Unbalanced HRV sat inside a green Stress & Recovery score (86). The report called the HRV out anyway, but the skill only requires recommendations for yellow and red domains.
    • Where a band is open-ended, counts whole days, or the rubric offers two methods, the exact score is the agent's judgment. For the HRV part of Stress & Recovery the rubric offers both Garmin's status and a percentage-below-baseline table, with no rule for which wins or which day's status to use; the demo used Garmin's status at the end of the period.
    • Garmin tools mix GMT and local timestamps, and the skill only warns about sleep start and end times. The agent has to work out which fields are local.
    • Irregular-rhythm and ECG alerts are checked only if your Garmin tools expose them; the server used in the demo does not.
    • The demo is a weekly report. Monthly mode uses the same rubric with sampled days and has not been shown here.
    • Not medical advice. It does not diagnose conditions and does not tell you to start, stop or change medication.
    • The demo below is one real run on Garmin data. It was checked against the source data afterwards: wording that went past the data was tightened and two breakdowns the skill requires (stress split, Body Battery drain) were added from the same data, with no score changed. Check key numbers in your own runs.

    This is wellness analysis of consumer wearable data, not a medical assessment: it does not diagnose, and its scores are heuristics for comparing you with yourself over time, not clinical grades. It needs a Garmin data source connected to your agent; without one it will not produce scores or trends. Wrist sensors, Body Battery and Garmin's stress score are estimates. Days without data are left out, and a metric missing for the whole period is skipped and reweighted; nothing is filled in with an estimate. Score bands are calibrated for active adults aged 18 to 30, and resting-HR bands have no age adjustment, so a clinically normal resting HR of 73 to 100 bpm scores yellow or red. Some judgments are still left to the agent: there is no rule for days the watch was only partly worn, no wear-coverage check before steps are scored (so low steps can partly reflect time off-wrist), and no rule for Garmin values that don't reconcile; the steps score does not account for sports such as cycling or swimming that log few steps; SKILL.md asks for bedtime and wake-time spread but the rubric's consistency bands are written for bedtime only; the two ways of scoring HRV have no tie-break; weekly reports have no previous-week baseline for sleep, resting HR or SpO2; and the overall light has no rule of its own beyond the per-domain bands. Irregular-rhythm and ECG alerts are checked only if your Garmin tools expose them. The demo shows a weekly report; monthly mode has not been demonstrated. The demo below is one real run on Garmin data. It was checked against the source data afterwards: wording that went past the data was tightened and two breakdowns the skill requires (stress split, Body Battery drain) were added from the same data, with no score changed. Check key numbers in your own runs.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Money-back guarantee

    Doesn't do what its page promises? You get your money back.

    Trust & safety

    Security scan

    Security scan passed 27 days ago

    Requires

    Garmin

    Permissions

    Write Files
    Read Files

    Allowed Hosts

    connect.garmin.com

    the report is built from your own wearable data, so the agent has to read it through the Garmin MCP server you have already connected. It only calls read tools, such as sleep, stress, heart-rate, HRV, Body Battery, SpO2, step and weight summaries, plus your profile for the age band and the saved sleep window. It never calls the write tools some servers expose (logging weight, blood pressure or food), makes no network requests of its own, and does not save or send your health data anywhere unless you ask. File reading is only for the skill's own scoring rubric and any earlier report you want it to compare against.

    Listed27 days ago

    What's inside

    Frequently Asked Questions

    No. It describes patterns in your wearable data and says when a pattern is worth asking a doctor about, naming the dates (for example SpO2 averaging below 94% on most nights, or resting HR up 15+ bpm for several days). It never names a condition and never advises on medication. If you mention chest pain, fainting, severe shortness of breath or similar symptoms, it stops the analysis and tells you to contact emergency services.

    A Garmin MCP server signed in to your own Garmin Connect account. The skill's tool names follow the open-source Taxuspt/garmin_mcp server, which the demo ran on (62 read-only calls, none failed). Other Garmin servers work if they offer the same functions; any function yours lacks is skipped and listed in the report.

    No. It only calls read tools and never the write tools some servers expose. It makes no network requests of its own and does not write your health data to files or other services unless you ask.

    Days with no data are left out of the averages and listed as missing; the skill never fills them with typical values. A metric with no data at all for the period, such as weight when you never weigh in, is skipped, its weight is redistributed, and the report says so. In the demo, sleep was scored on the 4 recorded nights, two partly recorded days were left out of the daily averages, and body composition was skipped because there were no weigh-ins. The skill does not yet have a rule for days the watch was worn only part of the day, so spotting them is left to the agent.

    Sleep, stress and heart scoring treat you like anyone else, and your exercise counts through intensity minutes. The steps score is the weak spot: it counts only steps, and rides or swims log almost none, so an athlete with plenty of training can still get a yellow Activity score. The demo shows exactly this. Training load and performance are out of scope; the skill leaves those to a training-analysis skill.

    New in Research & Analysis