# SEED-NOTES — life-helper seed extraction

Generated: 2026-06-12. Source-of-Truth files extracted into 8 JSON seeds at `seed-staging/`.

## Row counts

| File | Entries |
|---|---|
| `references.json` | 84 |
| `protocols.json` | 34 |
| `break_protocols.json` | 8 |
| `common_frames.json` | 5 |
| `methodologies.json` | 21 |
| `frame_patterns.json` | 30 |
| `reward_menu_items.json` | 30 |
| `diet_patterns.json` | 5 |

## Validation status

- Required-field violations: **0**
- `additionalProperties: false` violations: **0**
- Enum violations (RewardTier, FrameDomain, EvidenceStrength, huberman_fit_score 1..5, BreakCategory): **0**
- Orphaned `reference_ids[]` (pointing to non-existent reference): **0**
- Orphaned `default_common_frames[]` (break → common): **0**
- Orphaned `linked_protocol_ids[]` (diet pattern → protocol): **0**
- Orphaned `applicable_break_categories[]` (common → BreakCategory): **0**

## Unused references (5)

These references are valid but not yet cited from any entry because their hosting schemas
(`frame_pattern`, `reward_menu_item`) do not currently expose a `reference_ids[]` field.
They back the language-framing and reward-system theoretical foundation. Keep them — do
not prune.

- `ref_book_lakoff_elephant` — Lakoff (2004) *Don't Think of an Elephant!* → frame_patterns rationale
- `ref_doi_10_1037_0022_3514_53_1_5` — Wegner et al. (1987) White Bear / ironic process → frame_patterns + urge_surf
- `ref_doi_10_1037_0033_2909_125_6_627` — Deci, Koestner & Ryan (1999) overjustification meta → reward_menu_items
- `ref_doi_10_1146_annurev_psych_122414_033417` — Wood & Rünger (2016) habit conditioning → reward_menu_items
- `ref_doi_10_1207_s15326985ep3403_3` — Elliot (1999) approach vs avoidance → frame_patterns

## Decisions / notes

1. **Korean preserved verbatim.** Section symbols (§), em-dashes, full-width punctuation
   left as in source. No translation or normalization of body text.
2. **`omit` not `null`.** Missing optional fields are absent (not present with `null`).
3. **Methodology `is_core_engine: true`** applied only to `atomic_habits`,
   `tiny_habits`, `implementation_intentions` per project memory
   (`feedback_sustainable_minimal.md`).
4. **Break protocols** mapped the 4-week SoT structure into the schema's `phases[]`
   array with `week`, `goal`, `environment_design`, `if_then_examples`.
5. **`frame_patterns.json`** sourced from the "흔한 끊기 목표 변환 30선" table at
   `habit-todo-methodologies.md` lines 922–955. `domain` enum chosen from
   {food, drink, smoking, screen, porn, sleep, exercise, general}.
6. **`reward_menu_items.json`** sourced from "권장 리워드 메뉴 30선" lines 804–844.
   `tier_recommended` mapped T0–T4. `avoid_for_break_habits` populated where the
   reward involves caffeine/sugar/alcohol so it can collide with an active break habit.
7. **`diet_patterns.json`** kept to the 5 patterns explicitly enumerated in
   `diet-protocols.md` §2.1–§2.5. TRE/IF body in source is a stub pointer to §1.5;
   description was reconstructed from that section. `linked_protocol_ids[]` cross-links
   to the six `category: "diet"` protocols in `protocols.json`.
8. **No sections skipped.** All §1–§5 of every SoT are represented in some entry.
   §6 (한국어 해설자) of `diet-protocols.md` is intentionally not seeded — it is a
   user-fillable slot per project memory (`project_nutrition_module.md`).
9. **`reference_ids[]` ID style** used: `ref_doi_<doi-with-_>`, `ref_url_<slug>`,
   `ref_book_<author>_<slug>`, `ref_podcast_<show>_<ep>_<slug>`.