Home

UACRAgent Wiki

Welcome to the UACRAgent wiki.

UACRAgent is a desktop-first academic course assistant that helps students study from course materials using grounded AI support. It can chat over uploaded course documents and also generate structured study materials such as:

• review summaries
• practice booklets
• mock exams
• exam predictions

What UACRAgent Does

UACRAgent uses a retrieval-augmented generation (RAG) workflow:

1. You provide course materials such as syllabi, lecture notes, textbooks, assignments, and past exams.
2. UACRAgent loads and splits those materials into structured chunks.
3. It builds a local vector index for retrieval.
4. It uses the selected AI provider to answer questions or generate study materials grounded in those documents.

The project currently supports:

• Desktop GUI
• Interactive CLI
• Local FastAPI API

Main Features

• Persistent desktop study sessions
• Chat-based study assistance over course materials
• Document-type-aware chunking and retrieval
• Local Chroma vector storage
• Multiple AI providers:
  • Gemini
  • OpenAI
  • DeepSeek
• Multiple embedding options:
  • Gemini embeddings
  • OpenAI embeddings
  • Local embeddings
• Optional export to:
  • Markdown
  • DOCX
  • PDF

Platform Support

UACRAgent is designed to work on:

• macOS
• Windows
• Linux

The desktop GUI requires a Python environment with Tkinter support.

Local Embedding Notes

For source installs:

• local embeddings use the sentence-transformers path

For frozen standalone app builds:

• local embeddings use the ONNX local path
• in the current standalone build, only all-MiniLM-L6-v2 is supported for free local embeddings

Workspace Model

UACRAgent keeps its generated data in fixed app-managed locations.

Main storage areas include:

• Desktop auto-created sessions:
  • <app_data_dir>/sessions/<workspace_id>/
• CLI runs:
  • <app_data_dir>/cli_run/<workspace_id>/
• API runs:
  • <app_data_dir>/api_run/<workspace_id>/

Inside each workspace, UACRAgent stores its own files under:

• <workspace>/.uacragent/

Typical contents include:

• uploaded/cached file copies
• vector index
• outputs
• session state

Start Here

Recommended next pages:

• Getting Started
• Desktop Guide
• CLI Guide
• API Guide
• Workspaces and Data
• Privacy and Disclaimer
• Troubleshooting

Important Notes

• The API is intended for local trusted use
• Generated content should always be verified against official course materials and instructor guidance
• Third-party AI providers may process request data according to their own privacy policies and terms

License

UACRAgent is released under the MIT License.