agentleFS
kaustin923kaustin923/agent-fitness-coach/.claude/skills/daily-checkin/SKILL.md

daily-checkin

Skill0 starsChanged 4 months agoMITView on GitHub

Run the athlete's morning check-in — today's session card, verification of yesterday with a pace grade, flash insights, an energy check, and an optional pre-workout brief. Use when the athlete starts the day's first conversation, says good morning, asks "what's on today" or "what's my workout", reports training or a weigh-in in plain words ("I walked 2 miles", "185 this morning"), or wants a short brief before heading out.

1 other project uses a version of this.

Adopted by

Most recently taken up by

  • Other versions1 project uses a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 1

    1. Joshua-Palamuttam/workbench-template★ 3same name

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

  • In this repository12 other files it gives its agents.

The file

We split this skill into the rules it gives an agent

7 rules. About 0 are common, 4 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Do not run it as a substitute for weekly-review — the check-in never analyzes more than yesterday and today.

    line 16
  2. Distinctive:

    Ask the user: nothing up front.

    line 33
  3. Very specific:

    Freshen data if stale: If Strava (or a HealthKit export cadence) is the athlete's source and the last sync predates yesterday, run …

    line 41
  4. Very specific:

    Verify yesterday: Look up yesterday's row in plan.json and log.json:

    line 43
  5. Distinctive:

    Every total, percentage, and grade is computed with python and quoted verbatim.

    line 75
  6. Distinctive:

    Never say "Logged", "Done", or "Marked" unless you made the file edit this turn.

    line 76
  7. Distinctive:

    Keep the whole check-in tight.

    line 82

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Daily check-in

The daily check-in is the coach showing up before the athlete asks. It answers three questions in under thirty seconds of reading: what happened yesterday, what's on today, and is there anything worth knowing right now. Everything else waits for the weekly review.

When to run

  • The first conversation of the day, or any "good morning" / "what's today" / "what am I doing" message.
  • The athlete reports training in plain words — "I walked 2 miles", "did 40 minutes of yoga", "ran 5 this morning". Log it per step 9, even outside a full check-in.
  • As a scheduled morning routine (see guides/07-going-further.md for cron setup).
  • On request before a workout, for the compact pre-workout brief.
  • Do not run it as a substitute for weekly-review — the check-in never analyzes more than yesterday and today.

Inputs

Files read:

  • athlete/profile.md — units, surfacing preferences, check-in cadence, weight-prompt opt-in, goal.
  • athlete/plan.json — the active plan (status active); today's and yesterday's day rows.
  • athlete/log.json — done map, matched activities, fitness block with syncedDate.
  • athlete/macros.md — current targets, if nutrition coaching is set up.
  • athlete/food-log/YYYY-MM.md — today's rows, if any.
  • athlete/weight-log.csv — last 14 days.
  • athlete/goals.md — active goals.
  • athlete/state/insight-cooldowns.json and athlete/state/daily-insight.json — create either on first use.

MCP tools: none directly. If the athlete's data source is Strava and log.json's fitness.syncedDate is older than yesterday, run the strava-sync skill first (quietly) so verification works from fresh evidence. Tool names like list_activities belong to that skill; exact prefixes vary by harness (e.g. mcp__claude_ai_Strava__list_activities).

Ask the user: nothing up front. At most one question, and only when yesterday genuinely needs resolving.

Procedure

  1. Date ritual. Run date +%F for today, then build the weekday↔date table for this week and next with python (never in your head). Every weekday you mention below must be paired with its date from this table. Then check athlete/plan.json: if meta.race_date is within 21 days, offer (or run) the race-prep skill.
  2. Read state. Load the files above. If athlete/profile.md is missing entirely, offer the onboarding skill. If it exists but says status: in-progress, run the check-in from whatever files exist and offer to pick up the remaining setup questions afterward — an unfinished interview never blocks a good morning.
  3. Freshen data if stale. If Strava (or a HealthKit export cadence) is the athlete's source and the last sync predates yesterday, run strava-sync before verifying anything.
  4. Verify yesterday. Look up yesterday's row in plan.json and log.json:
    • Matched activity exists — one line with the actuals and the grade already recorded by sync: "Yesterday's tempo: 7.1 mi @ 8:38 — on target." Grade language and thresholds are defined in reference/tracking.md (## Run grading); quote the stored grade, never re-grade by eye.
    • Marked done but no activity — the athlete gets planned-distance credit: "Yesterday's easy 5 is logged as planned." (Credit rule: reference/tracking.md, ## Completion stats.)
    • Run day, not done, no activity — ask once, gently: "Yesterday's easy 5 isn't on Strava yet — did it happen?" If yes: set done[date] = true in athlete/log.json and the day's status to completed in athlete/plan.json this turn, then confirm. If missed: set status to skipped, no guilt, and note that plan-adjust exists if misses start stacking up.
    • Planned Walk day, not done — verify like a run day: ask once, credit the planned duration on a yes.
    • Rest or bonus cross-train day — skip verification; nothing to check. An unverified past day is a gentle nudge, never a failure state.
  5. Build the today card. From today's row in plan.json:
    • Lead line: "Today's session" when today has a workout; "Up next" plus the session's weekday and date when today is rest/open and the next session is later; "Your N-week build starts <weekday, date>" plus a week-1 preview before the plan begins; a countdown headline during race week; pure celebration and recovery guidance after the race.
    • Headline: use the day's title verbatim when present (it is authored in the canonical form, e.g. "Tempo · 5 mi @ threshold"). Otherwise construct by type: Race → "<meta.goal> · <distance>"; Rest → "Rest day"; Long → "Long run · N mi"; Tempo → "Tempo · {wo_miles} mi @ threshold"; Interval → "Intervals · " plus the workout text before the "@"; Marathon-Pace → "Marathon pace · {wo_miles} mi"; Easy/Recovery/Shakeout → "<Type> · N mi"; Cross-Train/Lift → the session's own content ("Upper-body strength"); Walk → "Walk · {duration_min} min (brisk)".
    • Body: structured runs (warmup non-null) get three lines — warm-up, workout, cool-down — plus total miles. Continuous runs get one "Target pace <range>/mi" line from paces[pace_zone]. A strength sidecar gets its own callout line.
    • Framing copy: rest days are productive — "Nothing to log — rest is training too. An easy walk or yoga is fine." Cross-train days connect to the week's purpose — "No running today — hit upper body and let the legs recover for Saturday's long run."
  6. Flash insights. Evaluate the decision table in reference/insight-rules.md (## Flash insights decision table) with python against the current local hour, today's food totals (servings-scaled, computed programmatically), macro targets, today's and this week's plan rows, the logging and training streaks, the recent weight-log rows, and the athlete's goal. Drop any rule that fired within its cooldown per athlete/state/insight-cooldowns.json; sort survivors by priority and surface at most 2. Write the fired rules' timestamps back to the cooldown file this turn.
  7. Daily insight (at most once per day). If athlete/state/daily-insight.json has attempted equal to today's date, skip — even if the earlier attempt produced nothing. Otherwise write attempted: <today> to the file before generating (the protocol lives in reference/insight-rules.md, ## Daily insight rotation — recording the attempt up front prevents silent retry loops), then pick the category by rotation, fall forward when a category's data floor isn't met, compute that category's statistics with python, and write 2–3 sentences following the style contract: lead with the pattern, use the exact numbers, never mention streaks or congratulations, never open with "Your". On success, fill in date, category, and text.
  8. Energy check. Only when at least one food item is logged today. Compute TDEE per reference/formulas.md (## Maintenance calories (TDEE)), classify consumed-vs-burned into a band (## Energy balance bands), and mention it directionally, not as a precise number. Time-of-day gate: before late afternoon (local hour < 16), frame intake as "so far today" and never as a deficit or surplus verdict — an 8 a.m. −1600 kcal "severe deficit" misleads; the day's energy direction belongs to the evening. If the athlete has logged zero food days in the last 7, say nothing about deficits or logging — adaptive engagement rules are in reference/insight-rules.md (## Proactive-message etiquette).
  9. Chat activity reports. When the athlete tells you what they did — "I walked 2 miles", "ran 5 this morning", "did 40 minutes on the bike" — that report is the data source; log it, don't just acknowledge it (a food report — "had eggs and toast" — belongs to the food-log skill instead):
    • Parse type, distance, and duration from the message; convert to storage units with python. If both distance and time are given for a run, compute the pace (reference/tracking.md, ## Pace math).
    • Upsert athlete/log.json: activities[date] = { "dist_mi": ..., "pace": ..., "name": "<short description>", "source": "chat" } (omit fields the athlete didn't give; never overwrite a Strava entry for the same date — device evidence wins).
    • If the report matches a planned day — same date, same kind of session (a run against a run day, a walk against a walking day, a lift against a Lift day) — set done[date] = true and the day's status to "completed" in athlete/plan.json, with actual carrying "source": "chat". Grade a run only when its pace is computable (## Run grading); otherwise leave the grade off.
    • Confirm in the athlete's units, in one line. A report that matches nothing planned is still training — it goes in the activities map and gets a friendly word, never a "that wasn't on the plan".
  10. Weight prompt. If the profile opts in and there is no entry for today: one line, easy to ignore. When a weight comes back, convert to kg, upsert athlete/weight-log.csv by date with source checkin (source chat when the athlete volunteers a weigh-in unprompted — "185 this morning"), and confirm in the athlete's units. A jump of 2 kg or more against the previous log gets a calm mention — day-to-day scale noise is normal, and the trend line (reference/formulas.md, ## Weight trend math) is what actually moves — plus an offer to recalculate macros via nutrition-setup when targets are on file.
  11. Pre-workout brief (on request). At most 280 characters, 2–3 sentences, no emojis, no greeting. Be specific to the session: a pacing or effort cue plus one practical tip — course, weather, or how this session sets up the next one — and a fueling note only when the athlete has logged at least one food item today and consumed less than 30% of BMR (an empty log at 6 a.m. is normal, not a crisis).

Rules

  • Every total, percentage, and grade is computed with python and quoted verbatim. Never sum food rows, derive paces, or total mileage mentally.
  • Never say "Logged", "Done", or "Marked" unless you made the file edit this turn.
  • Weekdays come from the date table, paired with their dates — no mental calendar arithmetic, ever.
  • Render all distances, paces, and weights in the athlete's declared units.
  • Grade language is neutral and three-valued: "Faster than target" / "On target" / "Slower than target". Slower is caution, never failure.
  • Missing data is a valid choice, not a deficiency. No guilt-trips over zeros, and an unchecked today never breaks either streak — the training streak counts from yesterday until today's session is confirmed, and the logging streak extends the same grace to a day whose food log hasn't started yet (reference/tracking.md, ## Adherence and streaks).
  • Respect the profile's surfacing preference: some athletes want three chat lines, some want the dashboard regenerated instead.
  • Keep the whole check-in tight. If the athlete asked a one-line question, answer in 1–3 sentences per COACH.md's length ladder and skip the ceremony.

Output

The today card, the yesterday one-liner, up to two insights, and the optional energy and weight lines — in that order, brief. If you ended by asking the athlete a question (e.g. resolving yesterday), stop there. Otherwise close with 2–3 natural next actions, such as:

  • "Log breakfast whenever you eat — I'll track macros from there."
  • "Tell me how it went afterward — 'ran 5, felt great' is enough and I'll log it."
  • "Want the 280-character brief before you head out?"
  • "Say the word and I'll regenerate your dashboard."

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.