START Project Documentation Hub
Welcome to the comprehensive documentation for the START (Scalable, Tailored Active-inference Research & Training) project - an advanced AI-powered system for creating personalized Active Inference and Free Energy Principle curricula.
๐ What is START?
START is a complete educational content generation pipeline that combines:
- Real-time research using Perplexity API for current domain insights
- Advanced LLM-based content generation via OpenRouter for professional-quality curricula
- Comprehensive personalization tailored to specific learners and professional domains
- Multilingual capabilities with full cultural adaptation
- Rich visualizations including charts, diagrams, and interactive elements
๐ Documentation Structure
Core Guides
๐ ๏ธ Setup & Development
- Environment Setup - Complete installation, configuration, and development guide
- Prerequisites, API setup, dependency management
- Development workflow, testing, troubleshooting
- IDE integration and advanced configuration
๐ System Architecture
- Pipeline Overview - Comprehensive system architecture and workflow
- acquire โ prepare โ process โ parse โ render curriculum pipeline
- Configuration-driven research approach
- API integration and content generation standards
๐งช Methods and Publication
๐ External Integrations
- Repository & Clone Management - External resource integration
- Active Inference Institute ecosystem integration
- Knowledge graph and implementation repositories
- Educational resource enhancement strategies
Specialized Documentation
๐ User Guides
๐ง Configuration Reference
- Entity Configuration:
data/config/entities.yaml - Target learner profiles
- Domain Configuration:
data/config/domains.yaml - Professional domain definitions
- Language Configuration:
data/config/languages.yaml - Translation targets
๐ฆ Data & Outputs
๐ Visualizations
๐ Translations
๐งญ Docs & Deployment
๐ Prompt Engineering
- Domain Analysis Templates:
data/prompts/research_domain_analysis.md
- Curriculum Generation Templates:
data/prompts/research_domain_curriculum.md
- Personalization Templates:
data/prompts/research_entity.md
- Translation Framework:
data/prompts/translation.md
๐ฏ Quick Start Paths
For New Users
- Environment Setup - Get up and running
- Pipeline Overview - Understand the system
- Usage Guide (GitHub)
For Developers
- Environment Setup - Development environment
- API Docs (GitHub)
- Tests (GitHub)
For Researchers
- Pipeline Overview - Research capabilities
- Clone Management - Access research repositories
- Configuration Files - Customize research targets
๐ External Resources
Active Inference Institute Ecosystem
Math & Programming Resources
In-Repo Entry Points
๐ System Capabilities
Research & Analysis
- Professional Domains: Life sciences, technology, business, healthcare, education, and whatever you prefer.
- Configurable Target Entities: Political figures, scientists, tech leaders, educators, and other explicitly configured audiences
- Live Research: Current industry insights and professional analysis when explicitly enabled
- Comprehensive Analysis: Prompt-targeted domain reports with explicit quality and evidence status
Content Generation
- Structured Curricula: Prompt-defined learning programs with manifest-backed quality and provenance
- Personalized Learning: Tailored strategies whose generated structure and evidence status are validated
- Modular Design: Prompt-defined learning units whose actual scope is recorded for review
- Assessment Integration: Built-in evaluation and progress tracking
- Data Visualizations: PNG charts with curriculum metrics and analysis
- Process Diagrams: Mermaid diagrams for structure and flow
- Interactive Elements: Visual learning aids and conceptual frameworks
Multilingual Support
- Configured Languages: Language and script mappings are loaded from
data/config/languages.yaml
- Cultural Adaptation: Full localization beyond literal translation
- Review Boundary: Structural and script checks are automated; language fluency and technical accuracy require human review
๐ง Configuration Overview
Research Configuration
# data/config/entities.yaml
entities:
- name: "karl_friston"
category: "scientist"
priority: "high"
# data/config/domains.yaml
domains:
- name: "biochemistry"
category: "life_sciences"
priority: "high"
Command-Line Interface
# From repository root; installed commands use the locked uv environment
uv sync --all-extras --dev
# Research high-priority entities
uv run start-curriculum --non-interactive --stages entity-research \
--entity-priority high --json
# Generate domain-specific curricula
uv run start-curriculum --non-interactive --stages domain-research \
--domains biochemistry --json
# Create multilingual content
uv run start-curriculum --non-interactive --stages translations \
--languages Spanish French --json
# Inspect generated output contracts and deterministic offline fixtures
uv run start-validate-outputs --check
uv run start-regenerate-offline --output-dir /tmp/start-fixtures --json
๐ Project Structure
START/
โโโ src/ # Core system implementation
โโโ learning/ # Curriculum creation scripts
โโโ data/ # Generated content and configuration
โโโ docs/ # Comprehensive documentation
โโโ tests/ # Test suite and validation
โโโ README.md # Project overview and quick start
๐ Development Workflow
Standard Development Cycle
- Configure targets in
data/config/ YAML files
- Run research using domain and entity scripts
- Generate curricula with comprehensive content creation
- Create visualizations for enhanced learning
- Translate content for multilingual accessibility
Quality Assurance
- Comprehensive testing with pytest and TDD approach
- Code quality with ruff linting and black formatting
- API integration testing for Perplexity and OpenRouter
- Content validation against Active Inference standards
๐ Getting Help
Documentation Resources
- This documentation hub for comprehensive guides
- Inline code documentation with detailed docstrings
- Example usage in test files and usage guides
- Configuration examples in YAML files
- Active Inference Institute for research questions
- GitHub Issues for technical problems and feature requests
- Test Suite for usage examples and validation patterns
START represents a new paradigm in educational content creation, combining cutting-edge AI research capabilities with comprehensive pedagogical design to produce world-class Active Inference curricula tailored to any professional domain or individual learner.
Institute Shortlinks (email-friendly)