BUD/S prep training: read today's workout, log tasks and PST results from your agent.
Scored across 12 tools
Most tools have distinct resource+action targets, and descriptions actively disambiguate the risky pairs (log_cardio/log_exercise log a value and mark done, while complete_task only resolves a task; the cooldown note clarifies when complete_task applies). The remaining soft spots are complete_task versus log_exercise for strength work and get_performance overlapping with get_today/list_pst_results on counts and PST history, but the descriptions steer an agent correctly.
Eleven of twelve names follow a clean snake_case verb_noun pattern (get_today, log_cardio, skip_task, list_pst_results), with a tidy skip/unskip pair. The only deviation is the bare verb 'upgrade', a minor blemish rather than a pattern break.
Twelve tools is squarely in the sweet spot for a training-plan domain, and every tool maps to a real user action: read today/week/performance, log cardio/strength/PST, resolve tasks, and upgrade. Nothing looks like padding or a redundant alias.
Read and write coverage of the daily loop is solid, but there are real dead ends: tasks can be skipped and unskipped yet never un-completed, and logging is hard-coded to 'today', so a missed session cannot be backfilled despite get_week showing history. Plan editing is also absent, which leaves the surface noticeably short of a full training lifecycle.
Mark one of today's tasks done: the cardio main task, the flexibility cooldown, or one strength exercise. The task id comes from get_today.
Inputs: task_id (required).
The performance page's numbers as text: the latest PST scored against the minimums, the PST history, and the plan's completion counts.
Inputs: no named parameters.
One PST result, scored against the BUD/S minimums.
Inputs: id (required).
Today's tasks from the user's active plan: the cardio and the strength exercises, each with its prescription and whether it is done, skipped or open. Every line carries the task id the write tools take.
Inputs: date.
The weekly training plan: one line per day with the cardio type, the strength type, and the day's status. Default is the current week.
Inputs: week.
The user's PST results, newest first, each scored against the BUD/S minimums (the same scoring the PST results page shows).
Inputs: no named parameters.
Log the time of today's cardio main task (the timed swim or run), as MM:SS or seconds, and mark it done. For an interval day, pass rep_number to log one interval's time.
Inputs: time (required), task_id (required), rep_number.
Log one set of a strength exercise (reps, and the weight in pounds; 0 for bodyweight), and mark the exercise done. Repeat the call for the next sets.
Inputs: reps (required), weight (required), task_id (required), notes, +2 more.
Log a PST result (the Navy Physical Screening Test): 500-yard swim, 2-minute push-ups, 2-minute sit-ups, max pull-ups, 1.5-mile run. The new result recalibrates the daily plan.
Inputs: situps (required), pullups (required), pushups (required), run_time (required), +3 more.
Mark one task skipped (a skip still counts as resolved for the day; the cooldown is optional and needs no skip). The task id comes from get_today.
Inputs: task_id (required), reason.
Undo a skip: the task returns to open. The task id comes from get_today.
Inputs: task_id (required).
Upgrade to Pro or Premium. Always two steps: without confirmPaymentSpend the tool only reads the exact price from Polar and charges nothing; call it again with confirmPaymentSpend: true to get the Polar checkout URL. The payment itself happens on Polar's page, never here.
Inputs: plan (required), confirmPaymentSpend.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Agent-first strength-training platform across iOS, Web & MCP: read & write workouts, PRs & plans.
iPhone strength log. Any MCP agent reads your training and plans workouts in the Tally app.
Workout journal Claude and ChatGPT write to: log workouts, get a program, see your progress.
Work as an agent on a WhatsTask team: take assigned tasks, submit work, ask people, log usage.
Add one secure layer between your agents and this server.
