---
name: doctrine-check
description: Audit a design choice or a code diff against Mike Mentzer's Heavy Duty doctrine. Use before committing a domain decision, or when a change touches training rules (intensity, volume, recovery, progression, substitution).
---

# Doctrine check

Audit the target against Heavy Duty doctrine and report the verdict.

Target (a design proposal, a diff, or both):

$ARGUMENTS

If the target is empty, audit the entire project, focusing on the core library.

## Source of truth

The doctrine lives in two places, in this order of authority:

1. `doc/Mike_Mentzer_Heavy_Duty_I.md` (HD1) — the primary source. It is
   copyrighted, gitignored, and never committed. Read it to cite exact wording.
2. `.kiro/steering/domain.md` — the distilled, binding principles. Use it when
   HD1 is not present.

Cite HD1 by section name. The sections are:

- I. Bodybuilders Are Confused!
- II. A Rational Approach
- III. The Science of Bodybuilding
  - The Intensity Factor
  - The Economics of Growth and Recovery
  - Conversations with Dorian Yates
- IV. The Ideal (or Principled) Routine
  - Ensuring Optimal Progress
  - Proper Exercise Performance
  - Training to Failure
  - Proper Weight Selection
  - The Pre-Stretch
  - Pre-Exhaustion
  - Negatives
  - Warming Up
  - Intensity and Muscular Function
  - Equipment
  - Regulating Volume and Frequency
  - The Routine
  - Important Points
- V. Individual Potential
  - The Role of Genetics
- VI. On Motivation
  - The Role of Values
  - Motivational Techniques

## The rule

On a conflict between the target and the doctrine, one of two outcomes holds:

1. The doctrine wins — the target must change, or
2. The conflict is raised explicitly.

Never let a conflict pass in silence. A silent deviation is the failure this
check exists to prevent.

## What to test against

Test the target against each binding principle. Report only the principles the
target touches; do not pad the report with untouched ones.

- **Intensity is categorical, not scalar.** A working set ends at momentary
  muscular failure; that is its definition (HD1, "The Intensity Factor": 0% at
  rest, 100% at failure). Effort past failure is qualitative
  (`Beyond_failure`: forced reps, negatives, rest-pause, static hold). Flag any
  numeric "intensity score".
- **Recovery produces growth; the gym only stimulates it.** Stimulation is
  immediate; production is not (HD1, "The Economics of Growth and Recovery").
  Recovery is systemic and time-based; a hard workout gates the next one.
- **Volume beyond the minimum is harmful.** One working set per exercise is
  nominal; three sets per muscle is the ceiling. The net negative of each extra
  set increases geometrically. Rising volume is a diagnostic warning, never an
  achievement or a target.
- **Progressive overload is the progress signal.** Progress is more reps,
  weight, or both; one extra rep counts. Reps are an outcome, never a target —
  a set ends at failure, not a number. The band only calibrates load. Raise the
  load 10-20% once twelve reps are reached. A 6-10 band is not breached at
  eleven.
- **Brief, infrequent, intense.** Shorter sessions at equal or better overload
  are favourable (HD1, "Regulating Volume and Frequency").
- **Focus over novelty.** Exercises come from a curated catalog. Substitutions
  follow author-specified whitelists; a routine may only narrow them.
- **On a stall, the answer is never more work.** After two weeks of no
  progress: a week off, then *reduce* volume and frequency — drop sets, add
  rest days (HD1, "Regulating Volume and Frequency"). No remedy may add volume
  or frequency.

## Encoding rules (for code targets)

- Prefer making an illegal training state unrepresentable over validating it at
  runtime.
- Deviations (training early, extra volume) must be possible but never silent:
  require an explicit acknowledgement and retain it.
- Keep plan and record distinct: prescription/routine vs logged workout;
  prescription vs progression.
- The logbook records; it never interprets. Invariants belong on authored
  plans, not on observed history. A log that refuses to state what happened is
  worse than one that records an error.

## Output

Report in this shape:

1. **Verdict** — one line: `Consistent`, `Conflict (doctrine wins)`, or
   `Conflict (raised)`.
2. **Findings** — for each principle the target touches: the principle, whether
   the target upholds or breaks it, and the HD1 section (or `domain.md`) that
   settles it.
3. **Required change** — for each `doctrine wins` conflict, the smallest change
   that restores consistency.
4. **Raised conflicts** — for each conflict the doctrine cannot settle alone,
   state the tension plainly and ask for a decision. Do not resolve it silently.

Keep it terse. Cite the doctrine for every claim. Do not invent doctrine that
is not in HD1 or `domain.md`; if the source is silent, say so.