Campus Copilot

A local-first academic decision workspace for students who want Canvas, Gradescope, EdStem, and MyUW in one structured place, then want clear answers to what changed, what matters first, and what to export or ask with cited AI.

Docs · Quickstart · Integrations · Distribution · Privacy · Product Brief · User Surfaces · Verification Matrix · Contributing · AI Collaboration · Security · License

Real workbench proof, not concept art:

Start Here In 60 Seconds

Campus Copilot takes four campus sites and turns them into one local workspace. The default student loop is:

1. sync Canvas, Gradescope, EdStem, and MyUW into one workbench
2. open the decision layer to see what changed, what is still open, and what to do first
3. export the same structured view or ask cited AI to explain it

That is the main story of the repo. It is not "open a blank chat box and hope the model figures school out for you."

Why This Exists

Campus Copilot is not a generic AI shell.

It is an academic decision workspace for students who want one place to answer questions like:

• What assignments are still open?
• What changed recently across my classes?
• What should I pay attention to first?

The product stays intentionally narrow:

• Structured data first: adapters normalize site-specific data into one shared schema.
• Local-first by default: storage, workbench views, filtering, and export live locally.
• AI after structure: AI can summarize or explain the workbench result, but it does not read raw DOM, raw HTML, or cookies.
• Export is a first-class feature: Markdown, CSV, JSON, and ICS are part of the core product, not an afterthought.

You can think of it like a school desk instead of a chat window: first gather the papers into one pile, then mark what changed, then ask for help on top of that organized pile.

What Changes After The First Sync

After the first real sync, the value is supposed to feel concrete:

• you stop hopping between four campus sites just to rebuild the same mental map
• Focus Queue, Weekly Load, and Change Journal tell you what changed and what should come first
• export presets carry the same structured evidence into Markdown, CSV, JSON, or ICS
• cited AI explains the same structured workspace instead of inventing its own hidden source of truth

What To Do First

If you are new, follow this order:

1. Understand the student-facing loop in this README.
2. Run the local workbench through Quickstart.
3. Read the product contract in docs/01-product-prd.md and docs/06-export-and-user-surfaces.md.
4. Only after that, use the proof and builder routes that match your intent.

That ordering matters. Proof is there to verify the workbench is real. Builder surfaces are there to consume the same substrate. Neither should replace the student-first story on the front door.

Pick The Right Entry Surface

If you already know why the product exists and just need the fastest truthful entry point, use this table instead of guessing:

| If you are | Start here | Current state | Quick proof | | :-- | :-- | :-- | :-- | | a GitHub visitor or reviewer | DISTRIBUTION.md and docs/storefront-assets.md | public front door is live | GitHub Pages returns HTTP 200 | | a student trying the product locally | Quickstart and docs/chrome-web-store-submission-packet.md | build-ready loc

... View full README on GitHub