training

/training — Generate today's plan

Build the training plan for today (or a given date).

Arguments

$ARGUMENTS Optional: date in YYYY-MM-DD format. Default: today.

---

Workflow

Step 1: Fetch athlete context

python3 "${CLAUDE_PLUGIN_ROOT:-.}"/scripts/fetch_context.py --date {DATE}
python3 "${CLAUDE_PLUGIN_ROOT:-.}"/scripts/log_muscle_load.py --backfill 7 --silent

The backfill ensures all activities of the last 7 days are processed in the muscle profile — even if new sessions came in since the last /analyse. Idempotent, fail-soft; never blocks planning.

Check dataWarnings. If present: 1 line to the athlete, then continue. Check athleteFeedback. If not "no feedback": show the feedback to the athlete (1 line) and make sure to pass it to the planner. Check skippedWorkouts. If present → inform the athlete and clean up: python3 "${CLAUDE_PLUGIN_ROOT:-.}"/scripts/delete_workouts.py --event-ids {IDs} Check hrvReviewPending. If present → before the planner ask the athlete (see CLAUDE.md "HRV readiness review"). Persist answer as NOTE, then continue with step 1.5.

Step 1.5: Exercise re-evaluation (CONDITIONAL — only when flagged)

Check planningConstraints for the 🔄 Exercise re-evaluation due flag (emitted by context_builder._compute_reeval_trigger at natural boundaries — recovery week, periodization phase change, or staleness).

• Flag absent (the normal case): skip this step entirely — the daily flow is unchanged. Do NOT run the reviewer; micro-progression via the specialists is enough.

• Flag present: re-challenge the exercise selection before the specialists carry it forward:

  1. Load a wider type history for the affected pillar(s) so the reviewer sees the real progression / sentiment trend:

     python3 "${CLAUDE_PLUGIN_ROOT:-.}"/scripts/fetch_type_history.py \
       --date {DATE} --type WeightTraining --tags {pillar tags} --max-sessions 8
     

  2. Launch the exercise-reviewer agent in a pane. Pass: the flag line (which trigger fired), the type history, and the date. The agent reads competition_plan.md, exercise_progressions.md (incl. the Re-Eval: blocks), and athlete_static.md itself.
  3. Present the reviewer's keep/progress/swap/retire recommendations to the athlete as one proposal (Coach decisiveness rule) and ask for confirmation. Never swap/retire silently — restrictions and standing prescriptions clear only by explicit athlete confirmation (see CLAUDE.md "Never silently drop or replace standing prescriptions").
  4. On confirmation, write the outcome back into config/exercise_progressions.md: update each reviewed exercise's Re-Eval: line — Status= (keep/progress/swap/retire) and letzte-Re-Eval={DATE}. This resets the staleness clock so the flag clears next session. Route any confirmed swap into the relevant specialist in step 3.