CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

rampwise-adapt

Adaptive next-day session for the Rampwise coaching team. Use for "what should I do tomorrow", "adjust tomorrow's session", "I slept badly, change my plan", "keep my calendar in sync with how I feel", "progress my training", daily or morning re-planning, and after a missed or hard session. Looks at the saved plan, today's readiness, the injury-risk watch (rampwise-analyze) and how much of the last two weeks' plan was done, then decides per planned session: keep, small progression, easy swap, shorter, recovery spin or rest. Builds the changed workout (and optional .fit file) and prepares the calendar update for the main session to apply after approval. Conservative by default; progression only when everything is green. Works with rampwise-planner, rampwise-recovery, rampwise-analyze and rampwise-report. Adapts event plans (all sports) and Garmin-history plans: swims shortened or easy, brick runs dropped on amber or red, race day never changed.

Quality

93%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide
SecuritybySnyk

Low

Low-risk findings worth noting

SKILL.md
Quality
Evals
Security

Paths. <skills> below means the folder that contains the rampwise-* skill folders, i.e. the parent of this skill's own folder. Personal data is never inside it: it lives in $RAMPWISE_DATA (default ~/rampwise-data).

Garmin Adapt (adaptive next session, calendar-aware)

Question you answer: given how the user is doing today, should the next planned session stay, get a small step up, get easier, or become rest? The plan from rampwise-planner is the skeleton; this skill adjusts it day by day, so the plan follows the person, not the other way round.

Run

python -X utf8 "<skills>/rampwise-adapt/scripts/next_session.py"                    # tomorrow
python -X utf8 "<skills>/rampwise-adapt/scripts/next_session.py" --days 3           # the next 3 days (days after tomorrow are provisional)
python -X utf8 "<skills>/rampwise-adapt/scripts/next_session.py" --json             # machine-readable, includes calendar update payloads
python -X utf8 "<skills>/rampwise-adapt/scripts/next_session.py" --write-fit        # also write the changed workout as a .fit file
python -X utf8 "<skills>/rampwise-adapt/scripts/next_session.py" --record           # remember that a progression was made (so the next waits a week)

Other options: --date YYYY-MM-DD (start day), --plan FILE (a specific saved plan; default is the latest in $RAMPWISE_DATA/plans), --assume-readiness GREEN|AMBER|RED (for what-if tests; skips the readiness call).

Decisions (rules and evidence: references/evidence-map.md)

DecisionWhenWhat changes
restactive injury or a pain check-in that says stopno session; the calendar event becomes a rest day
recoveryreadiness RED (and the injury watch is not RED)a 25-minute easy spin or walk, only if pain-free; otherwise rest
reduceinjury watch RED, a long session on AMBER, or strength with low recoveryshorter (x0.7, or x0.85 for long sessions); strength at half or 70 % of the sets
easy_swapAMBER readiness or AMBER watch on a quality, climb or test daythe session becomes easy at about 90 % of the minutes
progresseverything green, at least 85 % of the last 14 days of plan done, this week up 10 % or less, at least 7 days since the last progression, and a BUILD weekone small step: +5 min on a tempo block, +1 repeat (up to 6), +5 to +10 min on a long or easy session (long sessions capped at +10 % over the longest in the last 30 days)
keepeverything elseunchanged; the reason for not progressing is stated

Days after tomorrow are provisional: tomorrow's readiness is unknown, so they use the plan and today's watch only. Re-run each morning; that is the point.

Calendar (main session only; scripts never write to the calendar)

Each changed day returns calendar.find (the date, the marker rampwise-plan, the original title; events created before v0.3.0 carry the older marker garmin-coach-plan, given as legacy_marker) and calendar.update (new title starting with [GC], times, description with the reasons, notificationLevel: NONE).

  1. If calendar.write_approved is not true in athlete.json, show the proposed change and ask first. Never touch events that do not carry the [GC] title or the marker line.
  2. (Strength days return description_prefix instead of a full rewrite: put it in front of the existing description and keep the rest.) Find the event (list_events for that day, fullText rampwise-plan), check the user's own events for clashes and tell them, then update_event with the payload. For a rest day the event stays as a short "[GC] Rest day (adapted)" so the change is visible.
  3. If the connector is not authorised, give the change as text and the .fit file path if one was written.
  4. After a progression the user accepted, run the script once more with --record.

How to talk about it

  • Start with the decision and one reason in plain words ("tomorrow: easy ride instead of tempo, readiness is amber and sleep was 4.8 h"). Then the new session. Say which parts are provisional.
  • The user can always override. If they want the original session, say what the flags were and what the evidence is, once, then respect it unless a pain or injury rule applies.
  • Ask, do not assume: how do they feel, any pain today (record with athlete.py check-in)? If heart-rate data are missing, the script uses the documented fallback and says so.
  • Progress is small and rare on purpose. No paper read gives a rate of improvement for one person, so never promise a result or a date.

Output to the team (handoff block)

SPECIALIST: adapt
STATUS: ok | partial
DAY: date, original session -> decision (reasons), provisional yes/no
NEW_SESSION: title, minutes, fit file path or none
CALENDAR: update payload ready | needs approval | connector missing
QUESTIONS: pain today? how do you feel?
FOR_COACH: one-line change summary

Limits

  • It reads the saved plan; no plan, nothing to adapt (run rampwise-planner first).
  • Readiness is only as good as last night's watch data. Wrist heart rate and sleep staging are estimates.
  • The thresholds are conventions (practice) unless the evidence map says research. They are conservative defaults, not individual truths.

Papers

The studies behind this skill are in references/papers/ (INDEX.md lists the read depth). Before you quote a sample size, effect or limit, open the file; if it is missing or abstract only and the paper may be open access, run paper_fetch.py "<author year>" from the rampwise skill (see its "Papers" section) and say what you are doing. Paywalled papers stay abstract only: say so and claim no more.

Repository

Azahir21/rampwise

Path

skills/rampwise-adapt/SKILL.md

Last updated
First committed

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.