Execution System v1 — The Daily/Weekly/Quarterly Rhythm

Frame: Plans don't fail from bad content; they fail from missing infrastructure. Founder explicitly stated his bottleneck is "executing and staying consistent." This doc is the SYSTEM that turns the four plan docs into daily action, with Ray as the always-on adherence partner.

Principle: the lowest-friction logging surface wins. Founder already iMessages Ray dozens of times per day. Anything that requires opening a new app, filling a form, or remembering to log later will fail. Everything in this system rides on top of channels and tools the founder already uses.

1. Daily rhythm

Morning (passive, mostly auto-captured)

Action	How	Who/What
Wake weight	Apple Health → syncs via HAE → D1 daily_metrics. OR text Ray: "weighed 201.4"	Founder picks one. Apple Watch is the lowest friction if scale syncs already.
Sleep score	Apple Watch sleep tracking auto-syncs to HAE → D1	Zero action required.
HRV on wake	Apple Watch passive	Zero action required.
Morning brief delivery	iMessage from Ray ~6:27 AM ET	Includes HEALTH section (see section 2 below).

Throughout the day (active logging via iMessage)

Action	Trigger	Ray's response
Meal photo	Founder iMessages a photo of the plate. Optional 1-line caption (e.g. "lunch — chicken bowl").	Ray parses photo via vision, estimates kcal + protein + carbs + fat + fiber, logs the entry, returns 1-line ack with running daily totals.
Meal text	Founder describes meal in text instead of photo (e.g. "5 oz salmon, 1 cup rice, broccoli").	Same as above. Vision optional.
Water	Optional: text "3L done" or "halfway on water"	Ray logs, returns ack.
Workout	Strong app or Apple Workouts → auto-syncs via HAE → D1. No iMessage needed.	Ray reads D1 next morning.
Supplement taken	Optional: "supps done" or skip entirely (default-assume daily)	Track only if founder wants the streak counter.

Evening

Action	How
Wind-down cue	Optional: Ray can send a 10pm "wind-down" nudge if founder opts in (off by default — `feedback_no_babysitting` rule).
Sleep tracking	Apple Watch auto.

Friction principle: the founder takes a photo → Ray does everything else. Photo + 1-line ack is the entire active logging loop for the day. Estimated 30 sec total active time per meal × 3 meals = ~90 sec/day of explicit logging effort.

2. Morning brief HEALTH section (proposed addition to /morning-prep)

Below the existing brief structure, append a HEALTH section. Single line if everything's nominal, expanded if a nudge is warranted.

Format

📊 Health: weight 201.2 (-0.6 wk) · meals 3/3 logged 2,080 kcal 142g P · workout: Mon strength · sleep 7.1h (5d avg 6.8h)
   {Optional nudge line if a trend warrants it}

Components

Weight: yesterday's number + 7-day delta
Meals: count logged + total kcal + total protein
Workout: "yesterday" if completed OR "scheduled today" if today is a training day
Sleep: last night + 7-day average

Nudges (only when warranted)

The nudge line appears ONLY if a trend or threshold triggers it. Examples:

Trigger	Nudge line
Weight stalled 2 weeks, calories on target	"Weight stalled 2 weeks — consider 200 kcal cut OR add 15 min Zone-2/day. Reply to discuss."
3 consecutive sleep nights < 6 hrs	"Sleep < 6h 3 nights running. Recovery debt high; consider de-loading today's strength session."
Skipped scheduled workout 2 weeks running on same day	"Wed VO2max skipped 2 weeks. Move to Thu this week? Or drop one Zone-2 walk to make room."
Hit 7-day protein floor (140g+) 7/7 days	"7-day protein streak hit. Keep going."
Missed protein floor 3 days running	"Protein floor missed 3 days — Greek yogurt or protein shake addable to one meal/day?"

Tone discipline (feedback_no_babysitting + feedback_advisor_not_pair_programmer): never shame, never moralize, always offer a specific actionable adjustment. Trends are data, not character judgment.

Streak counter

A "consecutive days logged ≥ 1 meal" counter on the right side of the Health line:

📊 Health: weight 201.2 ... · streak 14d ✓

Reset to 0 on any day with zero meals logged. Founder sees the streak; misses aren't called out, but they reset the counter.

3. Weekly review (Sundays)

Either incorporated into /self-review or stood up as /health-review. Recommendation: bolt onto existing /self-review cron for now (Sunday morning), promote to standalone if it grows.

Weekly summary delivered to iMessage (Sunday 8am ET)

🩺 Weekly health review — week ending 2026-MM-DD

Weight: 201.2 → 200.4 (-0.8 lb), 4-wk trend -1.6 lb
Meals: 18/21 logged (86%), avg 2,090 kcal / 138g P / 35g fiber
Workouts: 4/4 scheduled completed (2 strength, 1 VO2max, 1 Z2 walk x4)
Sleep: avg 6.9h, 1 night <6h, 0 nights >8h
Supplements: not tracked

Recommendation: ON TRACK. Hold current plan another week.
   {Or: "ADJUST. {specific change}"}

Trigger thresholds for ADJUST (vs HOLD)

Trigger	Adjustment
Weight loss < 0.25 lb/wk averaged over 2 weeks AND calories on target	Cut 100 kcal/day OR add 15 min Zone-2/day
Weight loss > 1.5 lb/wk averaged over 2 weeks	Add 100 kcal/day (too aggressive; risks lean-mass loss)
Strength sessions missed 2+ weeks running	Re-evaluate scheduling: same day too tired? Move to morning?
Sleep average < 6.5h for 2 weeks	Move bedtime 30 min earlier; investigate cause (work? screens? stress?)
Gout flare occurs	Add water log requirement temporarily; review purine intake from last 3 days; trigger Dr Ambrose conversation if 2nd flare in 6 mo (allopurinol indication)
Protein floor missed 3+ days in a week	Auto-add 1 protein-anchor meal recommendation to next morning brief

Sunday actions for founder

Skim the weekly summary (60 sec)
If supplements running low: confirm Subscribe-and-Save scheduled OR re-order
If quarterly checkpoint due: see Quarterly section below

4. Quarterly cadence

Each quarter (every 90 days)

Action	When	What
Lab refresh	Q2 + Q4 each year, or quarterly if values are drifting	Re-run the [[2026-05-22-lab-order-list-v1]] panel. Update assessment.
Plan reassessment	After labs return	Issue plan v(n+1) reflecting new data. Adjust nutrition / workout / supplement.
Supplement stack audit	Quarterly	Drop anything without measurable individual benefit. Add anything new from labs (e.g. vitamin E if FibroScan shows NASH).
Dr Ambrose check-in	Annual minimum, quarterly if values active	Bring lab refresh + executable questions list.

Calendar anchoring

A calendar event already exists for Health Record Export quarterly on Aug 21 (per assessment doc context). Co-locate the lab-refresh request with that event so founder gets the prompt naturally.
Add new calendar event: "Health plan reassessment" recurring quarterly, 60 min Saturday morning, the week after labs return.

5. Meal-photo logging — infrastructure SPEC (build later, not tonight)

Today's reality (Day 1 implementation)

In-Ray-session logic only. No new infrastructure. When founder iMessages a meal photo:

Ray reads the photo (vision capability already exists in the session).
Ray estimates macros (calories / protein / carbs / fat / fiber).
Ray writes the entry to the running daily summary (in-context for today; serializable to a daily-summary file in ~/.claude/state/health/YYYY-MM-DD-meals.md if persistence needed).
Ray returns 1-line ack via iMessage.

Tradeoff: no cross-session memory by default. Each day starts fresh. Morning brief HEALTH section pulls from yesterday's ~/.claude/state/health/YYYY-MM-DD-meals.md file.

Effort: zero. Works tomorrow morning.

Promoted infrastructure (build IF usage justifies, ~30 min effort)

When daily meal logging is consistent for 2-3 weeks and the founder confirms he wants persistent searchable history, promote to D1:

Proposed D1 schema:

CREATE TABLE meals (
  id INTEGER PRIMARY KEY AUTOINCREMENT,
  date TEXT NOT NULL,            -- 'YYYY-MM-DD'
  time TEXT NOT NULL,            -- 'HH:MM'
  photo_url_r2 TEXT,              -- R2 URL to photo (nullable; text-only entries skip)
  estimated_calories INTEGER,
  estimated_protein_g REAL,
  estimated_carbs_g REAL,
  estimated_fat_g REAL,
  estimated_fiber_g REAL,
  foods_text TEXT,                -- comma-separated food list or freeform
  source TEXT DEFAULT 'imessage', -- 'imessage' | 'manual' | 'voice'
  recorded_at TEXT DEFAULT (datetime('now')),
  meal_type TEXT,                 -- 'breakfast' | 'lunch' | 'dinner' | 'snack' (nullable, inferred from time)
  notes TEXT
);
CREATE INDEX idx_meals_date ON meals(date);

Photo storage: new R2 bucket rdco-health-meal-photos (separate from existing rdco-health-data bucket for isolation).

Worker logic: ~50 lines of code to:

Receive photo URL + macro estimates from Ray session
Write row to D1
Return confirmation
Daily summary endpoint: GET /meals/daily/:date returns totals

Build trigger: when in-session logging hits a friction point (lost context, want trend search beyond 24 hrs, want to look up "what did I eat that day before the gout flare," etc.).

Recommendation: START WITH IN-SESSION LOGIC. Promote to Worker when justified. Avoid premature infrastructure.

6. Consistency-leverage points (designed against founder's stated bottleneck)

Founder's words: "I need help executing and staying consistent." The five design choices that address this directly:

1. Friction must be LOW

Photo + 1-line ack > form-filling > app-switching
Founder is ALREADY on iMessage all day. No new surface.
No "did you log today?" reminders by default (those become noise the founder learns to ignore).

2. Feedback must be IMMEDIATE

Macro estimate appears within seconds of the photo (Ray vision parse + ack)
Weight trend visible the next morning, not weekly
Workout impact shows up in morning brief next day
A 24-hr feedback loop catches drift before it becomes a 4-week drift

3. Streaks must be VISIBLE

"Consecutive days logged" counter in the morning brief HEALTH line
Founder sees compounding effort, not just isolated days
Streak resets quietly on a miss (no shame, just zero counter the next day)

4. Misses must be NORMALIZED

The morning brief NEVER shames a miss
Trends are shown as data: "weight up 0.4 lb, sleep 5.5h, no workout yesterday" — no judgment language
The nudge line offers a specific actionable adjustment, not a guilt trip
Bad week happens. Next morning brief shows the data and asks one thing.

5. Plan must ADJUST

The system proposes specific changes when data shows drift
Founder isn't expected to figure out interventions from raw numbers
Example: "Weight stalled 2 weeks → consider 200 kcal cut OR 15 min more Zone-2/day" is concrete. Not "you should reassess your approach."

7. Edge cases + escalation

Situation	Auto-response
Gout flare in progress	Founder texts "gout flare." Ray switches workout plan to upper-body only + adds water-log requirement + flags to schedule Dr Ambrose conversation if 2nd flare in 6 mo.
Travel week	Founder texts "traveling Tue-Fri." Ray switches plan: bodyweight strength substitution, expect more sleep variance, accept logging gaps. No nudges that week.
Skipped meal logging > 3 days	Morning brief omits HEALTH section entirely (no data to report). Re-engage when next meal photo arrives. No shame, no questions.
Weight up > 3 lb week-over-week without explanation	Likely water/sodium/stress; Ray surfaces in weekly review but does NOT alarm. Two weeks of trend before suggesting calorie adjustment.
Lab finding > Tier 2 (e.g. AST > 3x ULN)	Surface as Tier 3 escalation: "your AST result is X, which is above the threshold for routine. Suggest contacting Dr Ambrose this week — want me to draft the message?"
Founder explicitly opts out ("Ray, stop tracking food this week")	Pause meal logging, sleep tracking unchanged (passive), no nudges. Resume on his cue.

8. Pilot success criteria (8 weeks)

How we know this system works (or doesn't):

Metric	Success threshold	Failure threshold
Meal logging compliance	≥80% of meals logged	<50% by week 4 → reconsider friction
Weight loss trajectory	0.4-0.8 lb/wk average	<0.2 lb/wk avg by week 4 → reassess plan (likely calorie or activity adjustment)
Workouts completed	75%+ of scheduled	<50% → schedule restructure
Founder NPS (informal — does this feel sustainable?)	Yes	"Ugh, this is too much" → strip features
Plan iteration	v2 issued by week 12 with new labs	If labs not done → bottleneck is appointment scheduling, not the plan

If by week 8 the pilot is succeeding: continue, promote meal logging to D1 Worker, expand HEALTH section in brief.

If by week 8 the pilot is struggling: identify which of the 5 leverage points failed, simplify rather than add features. The system can shrink to "weight + workout + supplement reorder" if active food logging proves too much.

9. Cross-references

Plans this system executes: [[2026-05-22-nutrition-plan-v1]], [[2026-05-22-workout-plan-v1]], [[2026-05-22-supplement-plan-v1]]
Lab cadence quarterly: [[2026-05-22-lab-order-list-v1]]
Clinical context: [[2026-05-21-founder-health-assessment-v1]]
Morning brief skill being modified: ~/.claude/skills/morning-prep/SKILL.md
Existing related infrastructure: ~/rdco-vault/06-reference/2026-05-21-health-export-worker-architecture.md (if exists) + D1 rdco-health database
Existing self-review skill (where weekly review may bolt on): ~/.claude/skills/self-review/SKILL.md

10. What this system explicitly does NOT do

Does NOT moralize, shame, or "encourage" in cringey ways. Data + adjustment, that's it.
Does NOT require the founder to use a new app. iMessage + Apple Watch + Strong = his current tooling.
Does NOT create a new dashboard / web UI / Notion page. Morning brief HEALTH section is the only new surface.
Does NOT automate prescription decisions, labs, or Dr Ambrose communications. Founder + doctor own the clinical layer.
Does NOT measure things just because they're measurable. Each metric in the brief maps to a specific decision the founder might need to make that day or that week.

End of v1. Plan reassessment 8 weeks out (2026-07-17). If sustained, promote meal logging from in-session to D1+R2. If struggling, simplify before adding features.