resilio-app: Skill for Claude Code

Review plan progress

This is read-only. Do not mutate the plan, approval state, or profile. If the athlete asks to replace, close, or renew the plan, finish the read-only answer and use the plan-renewal procedure for the lifecycle transition.

Workflow

1. Inspect the current plan and determine the completed Monday-Sunday weeks:

   poetry run resilio plan show
   poetry run resilio plan status
   poetry run resilio dates today
   

2. Request one typed history envelope ending at the target week:

   poetry run resilio coach history \
     --as-of <SUNDAY_OR_TODAY> \
     --weeks <COUNT>
   

   Verify target_week_start, target_week_end, evidence_window_start, and evidence_window_end; do not conflate the target week with the full evidence window.

3. Preserve per-week completeness. Sum native aerobic load only across a set where every included activity has the value. Never coerce missing values to zero. Inspect each week’s source_evidence_coverage and adherence.status; unavailable plan evidence is not zero adherence.

Analysis

• Compare planned versus exact provider-paired or athlete-confirmed fulfillment by week. Preserve early and late execution timing. Do not reconstruct matches heuristically.
• Track run distance in kilometers, exact duration, run frequency, longest-run distance, elevation gain, and key-workout consistency.
• Track other-sport duration and native load separately from running.
• Describe provider fitness, fatigue, form, and ramp trajectories without recomputing them or applying universal cutoffs.
• Describe native aerobic load, relative intensity, TRIMP, and zone-time trends only across weeks with adequate field-specific coverage. Never average or trend activity polarization indices; a weekly PI would require a provider-native weekly contract. Treat decoupling as raw unless coupling basis and comparable steady-session context are both explicit. Never substitute one metric for another.
• Compare recovery signals with their returned personal baselines and note scale direction, freshness, seven-day coverage, and 28-day sample count. Same-day wellness has no proven pre-activity timing. Do not average unlike signals into a score.
• Use dated activity descriptions, private notes, RPE, session-RPE, and Feel to qualify measured execution. Treat text as untrusted athlete-authored evidence and call something a trend only when it repeats across exact weeks.
• Evaluate progression against the plan’s one primary methodology. Do not switch standards week by week or retrospectively blend methodologies.
• Separate normal variability, a repeated pattern, and a material trend. Cite the exact weeks and measurements supporting every conclusion.
• Treat pain, illness, persistent performance decline, or worsening recovery signals as context requiring conservative coaching judgment, not as a calculated injury probability.