Help

Click Sign In and authenticate with your Google account. First-time users are guided through a short onboarding flow that sets your sport, primary goal, and baseline training context.

Stryde works best with Apple Health data. Export your data from Apple Health (Profile → Export All Health Data) and upload the zip file via the Upload page. The more history you provide, the more accurate your Vitality, Fitness, and Readiness scores will be.

With 30+ days of heart rate and workout history, scores stabilize well. With 90+ days of data, the Vitality score's HRV and RHR trend components are fully meaningful. Fresh accounts with only a few days of data will see estimates labeled 'EST'.

On iPhone: open the Health app → tap your profile photo (top right) → Export All Health Data → Share → save the zip file. Upload that zip from Settings → Upload Data in the portal.

Yes. Generate an API key from the API Keys section on your Settings page. Use the key as a Bearer token in the Authorization header when posting to the upload endpoint. This is useful for automated nightly syncs from a trusted environment.

Apple Health export zip (export.xml inside), individual .xml files, and structured JSON for programmatic uploads. GPX and FIT file support is roadmap.

The ingest deduplicates records by source ID. If you re-upload data you already uploaded, the duplicate records are detected and skipped. This is expected behavior. Only truly new workouts increment the counter.

Three independent scores on a 1–100 scale. Vitality measures your long-term health trajectory (90 days). Fitness measures your current athletic capability (30 days). Readiness measures whether you should push hard today (today's state). See the full FAQ for scoring band definitions.

HRV suppression and elevated resting heart rate appear 24–48 hours before you consciously feel fatigued. Biomarkers are leading indicators. By the time you feel overtrained, you already are.

An EST (estimated) tag means the metric is using look-back logic because today's data hasn't arrived yet, or there are gaps in your history. Scores are still calculated; they're just derived from the most recent valid reading in the past 14 days.

Just say 'How am I today?' or 'What's my Readiness?' Claude returns your Vitality, Fitness, and Readiness scores plus your Ember & Flame status — a 2D quadrant that maps your fitness vs. fatigue into one of four states: INFERNO (fit and fresh — push), SMOLDERING (fit but fatigued — back off), FLASH (fresh but detrained — safe to work, limited ceiling), or ASH (fatigued and detrained — recover first). Each state ends with a single directive.

At session start, Claude surfaces any pending pattern-detected observations — things like 'your long run pace improved 4% over 6 weeks' or 'ground contact time rising 3 sessions in a row, possible form fatigue.' These fire when the system detects them. You don't need to ask.

Yes. At the end of every coaching session, Claude saves a summary to your account. When you return, it loads that context alongside your current metrics. You don't need to re-explain your goals or situation each time.

Yes. Say: 'I just did a 45-minute easy run, average HR 142, 7km.' Claude logs it directly. Include cadence, power, the shoe you wore, post-workout RPE — anything you know. Screenshots from your Garmin, Polar, Suunto, Coros, or Apple Watch work too — share the image and say 'log this workout.'

Just say it in conversation: 'My legs are wrecked today' or 'Energy maybe 4/10.' Claude infers and logs soreness, energy, mood, stress, motivation, and post-workout RPE — attaching the reading to today and linking it to the last workout if relevant. No form required.

Yes. Say: 'I slept 6h 40m, bed at 11pm, up at 5:40am.' Include deep sleep, REM, and core stage durations if you have them from your watch — they're factored into your Readiness score. Claude records the session and it flows into tomorrow's scores.

Yes. Stryde tracks running, cycling, walking, hiking, strength (upper/lower/full body/core), and HIIT. Strength sessions are tracked separately from aerobic load — they contribute to mechanical load, not CTL/ATL, so your training picture is accurate even if you're a cross-training athlete.

A weekly plan is a single week of scheduled sessions — easy run Monday, intervals Wednesday, long run Saturday. A program is a multi-week periodized structure: Base → Build → Peak → Taper → Race → Recovery. Programs are anchored to a race or goal date and auto-progress week by week. Claude picks up each phase where the last left off.

Tell Claude your goal and timeline: 'I'm running Boston Marathon on April 21. Build me a program.' Claude checks your current fitness level and injury risk first, then builds a phased program calibrated to where you actually are — not a generic 18-week template pulled from a book.

Yes. Say: 'I can't run Thursday this week, move it to Friday' or 'Shorten Saturday's long run by 20 minutes.' Claude shows a before/after preview and only applies the change when you confirm. Changes are version-controlled — Claude never silently overwrites your plan.

Tell Claude: 'I missed my Wednesday run.' It replans the remaining sessions in the week, accounting for the missed load — compressing what can safely fit or dropping what can't. It doesn't pretend the session happened, and it doesn't pile the missed work onto the next session blindly.

Yes. When generating a week, include your constraints: 'I can only train Monday, Wednesday, Saturday, Sunday' or 'I'm traveling Thursday through Sunday.' The week is built to fit your life while preserving the phase objective — base weeks stay base, peak weeks stay peak.

Say: 'I want to run a sub-48 Bolder Boulder 10K on June 1st.' Claude saves the race goal with your target time, distance, and date — then immediately runs a pace gap analysis showing your current best work pace versus your goal pace, and exactly how many seconds per km you need to find.

Yes. Provide your race venue altitude when creating the goal (e.g., 'the race is at 5,400 feet'). Claude applies a physiological altitude penalty to the pace target so your goal is grounded in what's actually achievable at elevation.

Within 3 weeks of race day, say: 'Build my race strategy for Bolder Boulder.' Claude generates a mile-by-mile pacing plan with HR ceilings per segment, fueling checkpoints, and specific risk warnings based on your known failure modes — things like going out too fast in mile 1 or cadence decay in the back half.

A Track is the goal container — it stores your race name, target date, and current periodization phase (Base/Build/Peak/Taper/Race/Recovery). A race goal stores the performance target (time, distance). They work together: the Track drives what kind of training to prescribe each week; the race goal drives pace analysis and race-day strategy.

Ask exactly that. Claude traces the causal chain: HRV suppression? ATL spike? Sleep debt? High ambient temperature? It cross-references your workout data against your baseline window and names the 2–3 most likely contributing factors with confidence levels — not just 'you were tired.'

Yes. Drift reports are computed every Sunday — improving / stable / declining / mixed trend across your training with a specific action item for the coming week. Claude also scans for positive breakthroughs: stride PRs, efficiency gains, aerobic gains at similar HR, and improved cadence under fatigue.

Ask: 'How does my HRV this month compare to February?' Claude runs a period comparison with delta, percent change, and a confidence level based on sample count. Works for any tracked metric: HRV, resting HR, sleep duration, sleep efficiency, workout distance, weight.

Ask: 'How did yesterday compare to my baseline?' Claude checks pace, power, HR, and GCT against your 90-day median and returns percentage deltas — so you know whether it was genuinely better or just felt better.

Ask: 'What are my PRs?' Claude returns your personal records across all activities: best 5K, best 10K, longest bike ride, longest hike, and strength bests — sourced from verified synced workouts only.

Yes, with Garmin, Polar, or compatible Apple Watch running dynamics. Claude tracks cadence (spm), ground contact time (GCT ms), stride length, and vertical oscillation across sessions. It plots your GCT trend over time — rising GCT across multiple sessions is an early sign of form fatigue or shoe breakdown.

Yes. For any structured interval session, Claude pulls work-period averages: pace, power, HR, GCT, cadence, rep count, pace consistency (CV), and whether power built or decayed across reps. This shows whether your intervals degraded, held, or built — three very different training stimuli with different implications.

Yes. Tell Claude: 'Add my Nike Vaporfly 3 — carbon-plate race shoe.' Claude registers the shoe with category, stack height, and retirement threshold (default 500km). Every run tagged to that shoe accumulates mileage. Claude warns you when you hit 90% of the threshold.

Yes, and Claude flags it. Carbon-plate shoes (Vaporfly, Alphafly, Endorphin Pro, etc.) provide mechanical assistance that can mask ankle and calf weakness over time. If all your recent sessions show carbon footwear, Claude notes potential motor-program interference and may suggest mixing in a daily trainer for easy runs to preserve unassisted mechanics.

Ask: 'Am I overdoing it?' Claude runs a full injury risk assessment: ACWR (acute:chronic workload ratio), training monotony, 7-day load ramp, sleep debt, and HRV trend direction. It returns a risk band — low / moderate / high / critical — with specific contributing factors and recommended actions tied to each one.

The Injury Check skill is a structured 'should I keep going?' workflow. It cross-references load, injury risk score, running mechanics, and ground contact time trend — and returns a single verdict: CLEAR (continue as planned), WATCH (reduce intensity, monitor), or BACK OFF (immediate recovery required), with the specific signals that drove it.

Claude can establish a return protocol: conservative ACWR ceilings, mandatory easy weeks, phase-locked intensity caps. The plan won't breach those limits even if you ask for more aggressive sessions — it tells you why, and what the progression timeline looks like instead.

Yes. For athletes 50+, the injury risk model includes age-stratified structural risk factors — patellofemoral and meniscal risk — that appear when load or HRV signals warrant. Claude flags these with higher urgency than generic muscle soreness warnings.

Four personas are available: Physiologist (evidence-based, data-forward — leads with metrics and research), Drill Sergeant (direct, high-accountability — short, expects follow-through), Longevity Coach (conservative load, health-first — prioritizes recovery and long-term structural health), and Endurance Veteran (tactical, race-focused — speaks in effort zones and execution). Say 'Switch me to the Longevity Coach' — you get a preview first.

Nothing changes. All workouts, plans, and scores transfer automatically. The previous coach's private athlete read is archived (not deleted) — if you return later, that context is still there. The new coach starts fresh with your data but builds its own understanding over time.

Yes. Say: 'Generate a coach handoff package.' Claude builds a signed, time-limited data package containing your patterns, current readiness, weekly plan, training protocol, and recent workouts. The package expires after 72 hours by default and is HMAC-signed — your coach can verify it hasn't been modified.

Skills are .md prompt files you paste into your Claude Project Instructions. They teach Claude a specific structured workflow — which Stryde tools to call, in what order, how to analyze the results, and how to format the output. Install once, trigger with a phrase. No need to explain the workflow each time.

Three skills are currently available: • Weekly Debrief — structured weekly review with GREEN / AMBER / RED signals across load, readiness, and plan execution. Locks in one specific focus for next week. • Race Prep — 3-phase taper strategy calibrated to how many days out your race is. Pulls fitness, load trend, and race goals for a concrete countdown plan. • Injury Check — 'am I overdoing it?' diagnosis. Returns a single CLEAR / WATCH / BACK OFF verdict with the signals that drove it. Download from the Skills page (/skills) in the portal.

Download the .md file → open Claude Desktop → go to your Project → paste the contents into Project Instructions (or drag the file in). Then use the trigger phrase in any new conversation. Example phrases: 'Run a weekly debrief' · 'Race prep, I'm 18 days out' · 'Run an injury check'.

Go to Settings → API Keys section. Delete the old key (Claude loses access immediately) and create a new one. Update the Authorization header in your Claude connector config with the new value.

Settings → Account → Delete Account. This triggers a cascade deletion of all your records from active systems. Deletion completes within 30 days except where legally required retention applies.

Stryde authentication is entirely delegated to Google. Recovering access requires recovering your Google account first. Contact Google account support. Once your Google account is recovered, sign in to Stryde normally.