athlete.get_context
ChatGPTAthlete context in one call: declared profile, Research-observed and inferred signals, and training state, each with provenance (calculated_at, data_quality, age_hours). Start here.
recommendation.get_next_action
ChatGPTNext training step from current training and recovery evidence, with limitations and a decision reference. Does not change training.
research.detect_plateau
ChatGPTPlateau status for a metric, with the evidence window and reason codes.
research.get_body_response
ChatGPTAssociation between the training pattern and body change. Not a causal claim.
research.get_correlations
ChatGPTWithin-user associations. These are not causal. Cite window, samples and data_quality.
research.get_insights
ChatGPTEvidence-backed findings already computed for this athlete. Cite window and data_quality.
research.get_metric_digest
ChatGPTOne metric over a window: n, first, last, delta, slope_per_week, coverage and data_quality. Use the on-screen domain, metric and window. A pending artifact is not a finding.
research.get_predictions
ChatGPTVersioned estimates with uncertainty. Do not present an ineligible prediction as current.
research.get_progress
ChatGPTProgress series for one domain: body composition, nutrition, wellness or executed training (not plan state). Includes coverage and data_quality.
research.get_timeline
ChatGPTPaginated athlete timeline.
research.list_hypotheses
ChatGPTRead-only hypothesis catalog and feature domains. Does not change a plan. Cite window, confidence and one limitation. These are associations, not causes.
training.confirm_plan
ChatGPTCreate exactly the previewed plan (use the token of the latest preview the user approved). Only after explicit approval; archives the active plan of the same sport.
training.confirm_session_adjustment
ChatGPTApply the proposed session adjustment. Only after the user explicitly approved it.
training.create_program
ChatGPTCreate exactly the previewed hybrid program (use the token of the latest preview_program the user approved). Only after explicit approval; both plans stay active.
training.explain_plan_decision
ChatGPTWhy a plan decision was made: strategies, eligibility, inputs and signal provenance.
training.get_current_plan
ChatGPTActive plan(s): goal_code, weeks, dates, available_days, chosen strategies and latest_decision_id. data.plans lists every active sport; pass sport to filter.
training.get_plan_overview
ChatGPTTraining state (data.sports), active plans (data.plans: strength and running may both be active) and today's sessions (data.sessions; empty with status none when there is no session) in one call. date and timezone select the session day; sport filters the reads. include_week adds the current week across all sports. Keep every returned plan and session.
training.get_today_session
ChatGPTPlanned or adjusted session(s) for a day (default today). data.sessions lists every sport; empty with status none when there is no session. Pass sport to filter.
training.list_plan_options
ChatGPTBefore creating a plan: goals, every strategy slot and option for the sport with eligibility, recommendation and reason codes, suggested weeks and start date. Relay only returned options.
training.preview_plan
ChatGPTPreview a complete plan for review before activation. Pass the conversation locale. Give a brief rationale and mention warnings that affect the choice; the app shows the prescription. summary.weeks_outline is authoritative: never invent exercises, sets, reps, loads or progression. For requested changes, preview again using returned summary.substitution_options slugs and the full customizations. Confirm only the latest preview the user approved.
training.preview_program
ChatGPTPreview coordinated strength and running plans before activation: exactly one of each sport. Pass the conversation locale. Explain the chosen priority and relevant schedule findings briefly; the app shows both prescriptions. summary.plans[] is authoritative; never invent prescriptions. Both plans stay unchanged until the user approves this preview.
training.propose_session_adjustment
ChatGPTPreview an adjustment with the actual before/after and assessment used. Read the session and plan versions first. Returns a confirmation_token only for an applicable change. Confirm only after the user reviews and approves this preview.