AGENTS.md — RegenBrief

Entry point for AI coding agents and automated tools working on this repository.

Project overview

This is a Jekyll static site — RegenBrief, provided by DYPHI Inc.’s Functional Longevity Center. A clinical archive and research knowledge base for Korean-speaking clinicians and researchers in regenerative medicine, frailty, and cognition.

Live site: https://flc.dyphi.com (or GitHub Pages staging)

Key files and directories

_clinical/          # Clinical trial entries (Markdown + YAML front matter)
_research/          # Research notes
_policy/            # Policy analysis
_data/
  indications.yml   # Controlled vocabulary for indications (id, name_ko, name_en, category)
  cell_sources.yml  # Cell source taxonomy (id, short_ko, family, origin_tissue)
  trials.yml        # ClinicalTrials.gov sync cache (auto-generated)
_layouts/
  default.html      # Base layout (header, nav, footer, Copy-for-AI button)
  clinical.html     # Clinical trial detail layout with JSON-LD
  note.html         # Research/policy note layout
_includes/
  clinical-card.html
  note-card.html
pages/              # Static index pages (clinical-archive, by-indication, etc.)
scripts/
  sync_clinicaltrials.py   # ClinicalTrials.gov API v2 sync script
assets/css/main.css
_config.yml
llms.txt            # AI agent site index
skill.md            # Capability declarations
robots.txt          # Crawler permissions
agent-permissions.json

Content schema — Clinical Archive

Front matter fields for _clinical/*.md:

Field	Type	Values
`layout`	string	`clinical`
`title`	string	Korean title
`date`	date	YYYY-MM-DD
`nct_id`	string	NCT##########
`sponsor`	string	Organization name
`sponsor_country`	string	ISO country name
`cell_source`	string	See `_data/cell_sources.yml` ids
`cell_origin`	string	`allogeneic` \| `autologous`
`indication`	string	See `_data/indications.yml` ids
`phase`	string	`phase-1` \| `phase-2` \| `phase-2a` \| `phase-2b` \| `phase-3`
`status`	string	`recruiting` \| `active` \| `completed` \| `terminated`
`delivery_route`	string	Korean description
`patient_count`	integer	Enrollment count
`result_summary`	string	One-sentence result summary (Korean)
`publication`	string	Citation or `"—"`
`publication_doi`	string	DOI without `https://doi.org/` prefix
`tags`	list	Lowercase, hyphenated

Development conventions

Language: Jekyll + Liquid templating; no JS build step
CSS: Single file at assets/css/main.css; CSS custom properties (--ink, --paper, --accent, etc.)
New clinical entry: Create _clinical/YYYY-slug.md following the front matter schema above
Data updates: _data/indications.yml and _data/cell_sources.yml for vocabulary extensions
Status sync: scripts/sync_clinicaltrials.py updates status/enrollment in _data/trials.yml via ClinicalTrials.gov API v2

Multilingual content — mandatory rule

Any time new content is added to a Korean collection, always create EN and RU translations in the same commit. This applies to all three content types:

RegenBrief (`_regenbrief/`)

Korean	English	Russian
`_regenbrief/`	`_regenbrief_en/`	`_regenbrief_ru/`
layout: `paper`	layout: `paper-en`, `lang: en`	layout: `paper-ru`, `lang: ru`

Translate: title, subtitle, key_summary, indication. Copy unchanged: tags, source_authors, source_title, source_journal, source_doi, source_year, date.

Clinical Archive (`_clinical/`)

Korean	English	Russian
`_clinical/`	`_clinical_en/`	`_clinical_ru/`
layout: `clinical`	layout: `clinical` + `lang: en`	layout: `clinical` + `lang: ru`

Translate: title, delivery_route, result_summary, body text and section headers. Copy unchanged: nct_id, sponsor, sponsor_country, cell_source, cell_origin, indication, phase, status, patient_count, publication, publication_doi, tags, date.

Section header mappings (KO → EN / RU):

## 임상 디자인 → “## Trial Design” / “## Дизайн исследования”
## 주요 결과 → “## Key Results” / “## Ключевые результаты”
## 임상적 의의 → “## Clinical Significance” / “## Клиническое значение”
## 한계 → “## Limitations” / “## Ограничения”
## 한국 시사점 → “## Korean Context” / “## Корейский контекст”
## 한국 도입 관점 → “## Korean Adoption Perspective” / “## Перспективы внедрения в Корее”
## 디자인의 특이점 → “## Design Notes” / “## Особенности дизайна”
## 후속 적응증 → “## Follow-on Indications” / “## Последующие показания”
## 후속 시험 → “## Follow-on Trials” / “## Последующие исследования”
## 관련 임상 → “## Related Trials” / “## Смежные исследования”

Policy / Stat (`_policy/`)

Korean	English	Russian
`_policy/`	`_policy_en/`	`_policy_ru/`
layout: `note`	layout: `note` + `lang: en`	layout: `note` + `lang: ru`

Translate: title, subtitle, category, all body text, table headers and cells. Copy unchanged: tags, date.

AEO (Agentic Engine Optimization) files

This site implements AEO. Do not remove or break these files:

File	Purpose
`robots.txt`	Explicit AI crawler allowlist
`llms.txt`	Structured content index for AI agents
`skill.md`	Capability and vocabulary declaration
`agent-permissions.json`	Machine-readable access rules
`pages/for-ai.md` (→ `/for-ai/`)	Clean markdown content index

Default layout (_layouts/default.html) includes:

<meta name="token-estimate"> — estimated token count for page
JSON-LD WebSite schema
“Copy for AI” button (copies clean article markdown to clipboard)

Clinical layout (_layouts/clinical.html) includes:

JSON-LD MedicalTrial schema with NCT ID, phase, sponsor, and status

Build and deployment

# Local development
bundle install
bundle exec jekyll serve

# ClinicalTrials.gov sync (manual)
python scripts/sync_clinicaltrials.py

# Production: GitHub Actions → GitHub Pages
# Workflow: push to main → auto deploy

Constraints and rate limits

ClinicalTrials.gov API v2: no authentication required; be polite with rate (~10 req/s)
No database; all content is flat Markdown + YAML
No JavaScript framework; keep the site buildable with standard jekyll-feed, jekyll-sitemap, jekyll-seo-tag plugins

ClinicalTrials.gov API v2: https://clinicaltrials.gov/data-api/api
Jekyll docs: https://jekyllrb.com/docs/
llms.txt spec: https://llmstxt.org/