CBT Programs Registry

The CBT Programs Registry makes CBT content easy to share, verify, and translate at scale.

It is designed for teams that want:

• reusable CBT programs for many health problems,
• transparent authorship and review status,
• fast multilingual expansion,
• easy integration into AI agent workflows.

Browse currently available programs in the CBT Program Viewer.

Why this is valuable

• Trust-first metadata: every program tracks aiCreated, humanCreated, humanReviewed, and healthcareExpertVerified.
• Easy localization: each language lives in locales/<lang>.json so translation is independent from core metadata.
• Safe collaboration: all updates are tracked in changelog.md and progress/*.md.
• Agent-native: SKILL.md + CLI commands support recommendation and guided support flows.

Program structure

Each program lives in:

skills/cbt-programs/data/programs/<program-id>/

Core files:

• program.json — metadata, lineage, recommendation signals, translation status
• locales/<lang>.json — language-specific content
• changelog.md — user-facing release log
• progress/*.md — implementation and review notes

Install

cd /path/to/eir-open
npm install -g ./skills/cbt-programs

Core workflows

List and recommend:

cbt-programs list --lang en
cbt-programs recommend --problem "difficulty sleeping" --lang en --limit 3

Create a new program:

cbt-programs scaffold cbt-shift-work-sleep \
  --title "CBT for Shift Work Sleep" \
  --condition "Shift Work Sleep Disturbance" \
  --creator ai-assisted

Improve an existing program:

cbt-programs improve cbt-shift-work-sleep \
  --lang en \
  --add-tags "sleep-hygiene,relapse-prevention" \
  --add-problems "night waking,rumination" \
  --mark-human-reviewed \
  --note "Improved symptom coverage and review state"

LLM-driven improve (host agent provides plan JSON):

cbt-programs improve cbt-shift-work-sleep \
  --llm \
  --llm-plan-file ./llm-improve-plan.json \
  --dry-run

Validate:

cbt-programs validate

Contributing new languages

This is intentionally lightweight:

1. Pick a program folder under data/programs/<id>/.
2. Add locales/<lang>.json (example: locales/es.json).
3. Add translation record in program.json.translations[]: lang, status, completion, path.
4. Keep status explicit: draft, human-reviewed, or expert-verified.
5. Run cbt-programs validate <id>.
6. Add a progress/*.md entry describing what was translated and reviewed.

This model lets many contributors work in parallel across languages without merge conflicts in core program metadata.

Example translation metadata entry in program.json:

{
  "lang": "es",
  "status": "draft",
  "completion": 45,
  "path": "locales/es.json"
}

Example locales/es.json starter:

{
  "language": "es",
  "title": "Titulo del programa",
  "summary": "Resumen del programa.",
  "modules": [],
  "aiSupport": {
    "coachingTone": "apoyo, practico, sin juicios",
    "systemPrompt": "Ayuda al usuario a aplicar un paso concreto por respuesta sin diagnosticar."
  }
}

LLM improve plan format

When using --llm, your host agent can provide a JSON plan via --llm-plan or --llm-plan-file.

Example plan:

{
  "programSummary": "Shorter, clearer summary for better adherence.",
  "addTags": ["relapse-prevention", "sleep-hygiene"],
  "addProblems": ["night waking", "bedtime inconsistency"],
  "newModule": {
    "title": "Relapse Prevention Toolkit",
    "overview": "Build a rapid recovery plan for setbacks.",
    "takeaways": ["Setbacks are expected and manageable."],
    "exercises": ["Create an early warning checklist."],
    "homework": ["Run a weekly relapse prevention review."]
  },
  "markHumanReviewed": true,
  "note": "Expanded relapse prevention and adherence support."
}

Contributing new programs

1. Use cbt-programs scaffold to create baseline files.
2. Fill in module content in locales/en.json (or your source language).
3. Set lineage and recommendation signals in program.json.
4. Add changelog and progress entries.
5. Run validation and open a PR.

Recommended PR checklist:

• Program validates with cbt-programs validate <id>
• Translation metadata is accurate (status, completion, path)
• Lineage flags are updated truthfully
• progress/*.md includes a clear change summary
• changelog.md includes a user-facing update note

Migrating legacy YAML

Import from the old CBT YAML folder:

node skills/cbt-programs/scripts/import_legacy_yaml.js \
  --source /Users/birger/Community/egen_journal/backend/content/programs \
  --target /Users/birger/Community/eir-open/skills/cbt-programs/data/programs \
  --status draft

Use --overwrite only when you intentionally want to replace existing program folders.