How harshan-164/edunexus works
The tree contains a substantial adaptive-learning product alongside the Agentic Slice Kit described by the README; the graph preserves both rather than treating the README as documentation for the whole tree. Nexora’s learner flow moves through diagnostic assessment, diagnosis, targeted remediation, and verification, with learner history informing the workflow. The slice-kit branch separately demonstrates reusable agent-run machinery. Edges are limited to excerpted calls, indexed references, or README statements; several frontend-to-backend and unsampled integration links therefore remain omitted.
Main entry points: workflow.py Learning Workflow · App.jsx Web App Shell · diagnostic.py Diagnostic Agent · runner.py Run State Machine
Components 27
Learner Experience
- Web App Shell
frontend/src/App.jsx - Learning Pages
frontend/src/pages - Progress View
frontend/src/pages/Progress.jsx - Application API
server.py - Streamlit Learning UI
app.py
Adaptive Learning
- Learning Workflow
backend/graph/workflow.py - Diagnostic Agent
backend/agents/diagnostic.py - Misconception Diagnosis
backend/agents/diagnostic.py - Remediation Agent
backend/agents/remediation.py - Verification Agent
backend/agents/verification.py - Revision Agent
backend/agents/revision.py - Agent Tools
backend/tools/agent_tools.py - LLM Adapter
backend/agents/llm_utils.py - Learning Contracts
backend/models/schemas.py
Learner Data & Media
- Learner History
backend/memory/learner_memory.py - Syllabus Retrieval
backend/memory/syllabus_memory.py - Account Data Services
backend/auth/user_services.py - Lesson Media
backend/media/lesson_media.py
Agentic Slice Kit
- Run State Machine
slice/runner.py - Append-Only Store
slice/store.py - Model Calls
slice/llm.py - Attempt & Token Fences
slice/budget.py - Runtime Config
slice/config.py - Vector Retrieval
slice/retrieve.py - Human Callback
slice/callback.py - Expert Form
web/expert.py
Other components
- Learner
Connections 27
- Learner → Web App Shell: uses
frontend/src/App.jsx - Web App Shell → Learning Pages: renders
frontend/src/App.jsx - Web App Shell → Progress View: renders
frontend/src/App.jsx - Web App Shell → Application API: fetches learner summary
frontend/src/App.jsx - Learner → Streamlit Learning UI: uses
app.py - Streamlit Learning UI → Learning Workflow: reads workflow state
app.py - Learning Workflow → Diagnostic Agent: generates questions
backend/graph/workflow.py - Learning Workflow → Misconception Diagnosis: requests diagnosis
backend/graph/workflow.py - Learning Workflow → Remediation Agent: requests intervention
backend/graph/workflow.py - Learning Workflow → Verification Agent: routes verification
backend/graph/workflow.py - Learning Workflow → Learner History: loads context
backend/graph/workflow.py - Learning Workflow → Learner History: records attempts
backend/graph/workflow.py - Diagnostic Agent → Agent Tools: uses tools
backend/agents/diagnostic.py - Diagnostic Agent → Learning Contracts: validates response
backend/agents/diagnostic.py - Diagnostic Agent → LLM Adapter: calls completion
backend/agents/diagnostic.py - Remediation Agent → LLM Adapter: calls completion
backend/agents/remediation.py - Verification Agent → LLM Adapter: calls completion
backend/agents/verification.py - Revision Agent → Lesson Media: uses lesson media
backend/agents/revision.py - Account Data Services → Learner History: provides service
backend/auth/user_services.py - Account Data Services → Syllabus Retrieval: provides service
backend/auth/user_services.py - Run State Machine → Append-Only Store: persists runs
- Run State Machine → Model Calls: runs model steps
- Model Calls → Attempt & Token Fences: enforces limits
slice/llm.py - Model Calls → Runtime Config: loads settings
slice/llm.py - Vector Retrieval → Append-Only Store: searches same database
slice/retrieve.py - Run State Machine → Human Callback: supports suspension
- Expert Form → Human Callback: answers callback
web/expert.py