An AI-powered curriculum generation tool for peer-led learning cohorts. Rhea uses Claude AI with optional web research to generate comprehensive, up-to-date module specifications and complete multi-week course structures and module specifications.
Rhea contains multiple workflows & utilities for curriculum generation and management:
| Themis | Tethys | Metis | Theia |
|---|---|---|---|
 |
 |
 |
 |
| Curriculum Creation | Arc Creation | Module Creation | Import, Export & Preview |
Rhea can use web search to ensure your curriculum reflects current industry practices and technologies, not just what Claude knew at training time.
Themis: complete multi-week courses
- Thematic arcs organizing related modules
- Learning progression across modules
- Course-level narratives and structure
- Individual module generation (coming soon)
Tethys: Arc generation and management tools (planned)
- Standalone arc creation between course and module levels
- Arc-level learning progression design
- Thematic coherence across related modules
Metis: standalone module specifications
- Learning objectives and module overview
- Detailed project briefs with examples and success criteria
- Research topics with guidance for learners
- Additional skills categorized by importance
- Project "twists" to add interesting challenges
Theia: content management and export
- Upload previously generated course JSON files
- Resume workflows from where you left off
- Export to Markdown, HTML, or JSON formats
- Configurable detail levels and section selection
- Round-trip capability: export → upload → continue
- Mnemosyne: long-term storage and versioning
- Atlas: Atomic Learning Assembler
- Node.js 20+
- Anthropic API key (get one here)
# Clone repository
git clone https://github.com/foundersandcoders/rhea.git
cd rhea
# Install dependencies
npm install
# Configure environment
cp .env.example .env
# Edit .env and add: ANTHROPIC_API_KEY=sk-ant-api03-xxxxx
# Run development server
npm run devOpen http://localhost:5173 to start using Rhea.
See Getting Started Guide for detailed installation and usage instructions.
- Claude Sonnet 4.5 pricing: ~£2.20/million input tokens, ~£11/million output tokens
- Typical module generation: £0.15-£0.30 (with research enabled)
- Course structure generation: £0.10-£0.20
- New Anthropic accounts often include free credits
Complete multi-week courses
- Configure course identity and logistics
- Plan thematic arcs
- Organize modules within arcs
- AI generates detailed structure
- Review and refine
- Export course overview or structure (JSON for re-upload)
- Generate individual modules (coming soon)
Note
Coming soon
Quick standalone modules
- Upload XML inputs (projects, skills, research)
- Provide structured context
- Generate with optional research
- Export preview or download XML specification
Theia: Content Management & Export
- Upload & Resume: Upload JSON course files to continue workflows in Themis
- Round-trip capability: Export → upload → continue working seamlessly
- Multiple formats: Markdown, HTML, JSON (PDF planned)
- Flexible detail levels: Minimal, summary, detailed, or complete
- Selective exports: Choose specific sections to export
- Course or module exports: Works with both Metis and Themis outputs
- Table of contents: Optional navigation for longer exports
- Drag-and-drop interface: Easy file upload with validation
<p>Export at any stage, then re-upload to continue work later. Upload existing course structures to review, refine, or generate additional modules.</p>
📝 Change Tracking & Provenance
Every generated module includes comprehensive change tracking to support the cascade pattern — where AI-generated modules are updated iteratively whilst maintaining human oversight:
<ul>
<li><strong>Automatic Changelog</strong>: Documents what changed, why, and with what confidence level</li>
<li><strong>Confidence Scoring</strong>: High/medium/low confidence flags help reviewers prioritize</li>
<li><strong>Research Citations</strong>: Web research sources automatically cited</li>
<li><strong>Provenance Tracking</strong>: Shows when generated, by which model, and flags sections needing review</li>
</ul>
<p>This enables curriculum councils to:</p>
<ul>
<li>Quickly identify what's been updated since last version</li>
<li>Focus review time on low-confidence changes</li>
<li>Understand rationale behind AI-proposed updates</li>
<li>Track sections needing human review</li>
</ul>
<p>See <a href="/docs/dev/work-records/changelog-schema-design.md">Changelog Schema Design</a> for technical details.</p>
🔬 Deep Research Capability
Enable web search during generation to:
- Verify technologies/practices are current
- Update recommendations based on industry trends
- Search trusted domains: vendor docs, GitHub, Stack Overflow, academic sources
- Cite sources for transparency
<p><strong>This is the core value</strong> — without research, you're just reformatting existing content.</p>
✅ Schema Validation
All generated modules are automatically validated against requirements:
- Minimum 3 module objectives
- Minimum 5 primary research topics
- Minimum 2 project briefs with detailed criteria
- Automatic retry (up to 3 attempts) if validation fails
- Getting Started - Installation, setup, first module
- About Rhea - What it does and why it exists
- Executive Summary - Quick overview
- Technical Overview - Architecture, components, patterns
- Architecture Decisions - Historical context and rationale
- Changelog Schema Design - Change tracking implementation
Metis accepts XML files for module inputs. Each file must have a specific root element:
projects.xml
<Projects>
<ProjectBriefs>
<ProjectBrief>
<Overview>
<Name>Project Name</Name>
<Task>What learners will build</Task>
<Focus>Key technologies and techniques</Focus>
</Overview>
<Criteria>Success criteria as bullet points</Criteria>
<Skills>
<Skill>
<Name>Skill Name</Name>
<Details>What learners will learn</Details>
</Skill>
</Skills>
<Examples>
<Example>
<Name>Example Name</Name>
<Description>Brief description</Description>
</Example>
</Examples>
</ProjectBrief>
</ProjectBriefs>
</Projects>
<p><strong>Minimal valid:</strong></p>
<pre><Projects></Projects></pre>
skills.xml
<AdditionalSkills>
<SkillsCategory>
<Name>Category Name</Name>
<Skill>
<SkillName>Specific Skill</SkillName>
<Importance>Recommended / Stretch / Essential</Importance>
<SkillDescription>Brief description</SkillDescription>
</Skill>
</SkillsCategory>
</AdditionalSkills>
<p><strong>Minimal valid:</strong></p>
<pre><Skills></Skills></pre>
<p>or</p>
<pre><AdditionalSkills></AdditionalSkills></pre>
research.xml
<ResearchTopics>
<PrimaryTopics>
<PrimaryTopic>
<TopicName>Topic Name</TopicName>
<TopicDescription>What to research and how to approach it</TopicDescription>
</PrimaryTopic>
</PrimaryTopics>
</ResearchTopics>
<p><strong>Minimal valid:</strong></p>
<pre><ResearchTopics></ResearchTopics></pre>
See Getting Started Guide for common XML gotchas and detailed format specifications.
- SvelteKit - Application framework with SSR and API routes
- Node.js 20+ - Runtime environment
- LangChain + Claude - AI orchestration (Claude Sonnet 4.5)
- Zod - Runtime validation and type safety
- XML - Curriculum data format
See Technical Overview for architecture details.
# Install dependencies
npm install
# Run development server
npm run dev
# Build for production
npm run build
# Preview production build
npm run previewRhea can deploy to:
- Vercel
- Netlify
- Node.js servers
- Other platforms (see SvelteKit adapters)
Important: Set ANTHROPIC_API_KEY in your hosting platform's environment variables.
Issues and pull requests welcome! Check the roadmaps for current priorities.
Built for peer-led learning communities.
Need help? See the Getting Started Guide or create an issue.
