How darl-genai/instructional_agents works
Instructional Agents generates and evaluates course materials through an ADDIE-based multi-agent workflow, with web, REST API, and CLI entry points. Users provide course requirements and optionally a catalog or feedback; the system generates instructional content and slides, then compiles or converts the results. A separate optimization workflow analyzes existing PDF slides and creates improved materials. The README also documents catalog and copilot modes, progress monitoring, evaluation, and LaTeX/PPTX outputs. Edges reflect README statements and shown source references; unsampled API routes and internal evaluation wiring are not inferred.
Main entry points: ADDIE.py ADDIE Workflow · ADDIE_optimize.py Optimization Workflow · run.py CLI · api_server.py REST API
Components 19
User Interfaces
- Web Interface
frontend/app.js - REST API
api_server.py - CLI
run.py - Catalog Inputs
catalog
Course Generation
- ADDIE Workflow
src/ADDIE.py - Agent Deliberation
src/agents.py - Slide Authoring
src/slides.py - OpenAI LLM
Slide Optimization
- Optimization Workflow
src/ADDIE_optimize.py - PDF Slide Processing
src/pdf_processor.py - Slide Knowledge Base
src/slide_knowledge_base.py - Slide Improvement
src/optimize.py
Evaluation and Output
- Course Evaluation
src/evaluate.py - Evaluation Entry
evaluate.py - LaTeX Compilation
src/compile.py - PPTX Conversion
src/latex_to_pptx.py - PPTX Builder
src/build_pptx.js - Course Materials
Other components
- Course Designer
Connections 24
- Course Designer → Web Interface: submits requirements
README.md - Course Designer → CLI: starts generation
README.md - Web Interface → REST API: requests tasks
README.md - REST API → CLI: invokes workflow
api_server.py - CLI → ADDIE Workflow: runs generation
run.py - CLI → Optimization Workflow: runs optimization
run.py - Catalog Inputs → ADDIE Workflow: guides generation
README.md - ADDIE Workflow → Agent Deliberation: creates deliberations
src/ADDIE.py - ADDIE Workflow → Slide Authoring: creates slide decks
src/ADDIE.py - Agent Deliberation → OpenAI LLM: requests completions
src/agents.py - Optimization Workflow → PDF Slide Processing: extracts slide content
src/ADDIE_optimize.py - Optimization Workflow → Slide Knowledge Base: builds slide index
src/ADDIE_optimize.py - Optimization Workflow → Slide Improvement: optimizes chapters
src/ADDIE_optimize.py - Slide Improvement → Slide Knowledge Base: searches context
src/optimize.py - Slide Improvement → Agent Deliberation: runs deliberation
src/optimize.py - ADDIE Workflow → LaTeX Compilation: compiles LaTeX
src/ADDIE.py - LaTeX Compilation → PPTX Conversion: offers conversion
src/compile.py - REST API → PPTX Conversion: converts task slides
api_server.py - PPTX Conversion → PPTX Builder: builds presentation
src/latex_to_pptx.py - ADDIE Workflow → Course Evaluation: evaluates materials
- Evaluation Entry → Course Evaluation: uses evaluator
- LaTeX Compilation → Course Materials: produces PDFs
- PPTX Builder → Course Materials: produces PPTX
- Course Evaluation → Course Materials: assesses materials