StudyPlanner

Course planning app for Computer Science students at the University of Tübingen.

• Web app and public AI gateway: https://studyplaner.pages.dev
• Public AI/OpenAPI metadata: https://studyplaner.pages.dev/api/ai/meta
• Public MCP endpoint: https://studyplaner.pages.dev/mcp
• Privacy policy: https://studyplaner.pages.dev/privacy

AI integrations

The public AI integration is read-only and unauthenticated. It exposes course catalog search, course-reference resolution, and course detail lookup only. It does not expose StudyPlanner accounts, profiles, progress, semester plans, transcript data, passwords, browser tokens, OpenAI keys, or Anthropic keys.

ChatGPT Custom GPT setup

1. Open ChatGPT → Explore GPTs → Create.

2. Configure → Actions → Import from URL.

3. Use this OpenAPI URL:

   https://studyplaner.pages.dev/api/ai/openapi.json
   

4. Authentication: None.

5. Privacy policy URL:

   https://studyplaner.pages.dev/privacy
   

6. Suggested GPT instructions:

   You help Informatik students at the University of Tübingen find courses in the StudyPlanner catalog. Use searchCourses for course searches, resolveCourse when the user gives a course number, and getCourseDetail before making claims about a specific course. Always mention the course number and title when available. The integration is read-only and has no access to personal StudyPlanner data.
   

7. Test prompts:

   • Welche Machine-Learning-Kurse gibt es?
   • Zeig mir Details zum passendsten Kurs.
   • Suche Seminare mit 6 ECTS im Sommer.

Claude / MCP setup

Use the public MCP endpoint:

https://studyplaner.pages.dev/mcp

If your Claude or MCP client supports remote MCP/connectors directly:

1. Add a custom MCP connector/server.
2. Name: StudyPlanner.
3. URL: https://studyplaner.pages.dev/mcp.
4. Authentication: None.
5. Test with: Nutze StudyPlanner und suche Kurse zu Machine Learning.

Use the exact /mcp URL. If Claude shows Dieser Konnektor hat keine Tools verfügbar, remove the connector and add it again after deployment; Claude can cache failed discovery attempts.

For Claude Desktop clients that still need a local bridge, add this to claude_desktop_config.json:

{
  "mcpServers": {
    "studyplanner": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-remote",
        "https://studyplaner.pages.dev/mcp"
      ]
    }
  }
}

Then restart Claude Desktop. Older clients that ask for an SSE discovery URL can try:

https://studyplaner.pages.dev/sse

Development

• Frontend: React 19, Vite, Tailwind CSS 4
• Backend: Python Cloudflare Worker
• Database: Cloudflare D1 / SQLite migrations
• Catalog imports: local Python tooling for ALMA data

Checks before committing

npm run test:frontend
cd frontend
npm run lint
npm run build

When changing transcript parsing, also run:

cd frontend
npm run validate:transcripts

Runtime guardrails

• Active D1 binding: DB
• Active D1 database: studyplanner-db (80ca9092-ddc6-454a-b04a-8ccae85ef2f5)
• Previous test DB: studyplaner-db-test (297f7a28-9069-431d-b989-49acf2537513)

Do not change the active database binding without explicit approval. Never commit AUTH_TOKEN_SECRET or generated secrets.

Useful docs:

• docs/ai-integrations-setup.md
• docs/cloudflare-runtime-config.md
• docs/cloudflare-development.md
• docs/authentication.md
• docs/mobile-testing.md
• backend/README.md

Quick local commands

Local development:

# terminal 1: frontend dev server (http://localhost:5173)
cd .\frontend\
npm run dev

# terminal 2: local backend Worker (http://localhost:8787)
cd .\backend\
npx wrangler dev

Local public AI gateway smoke test:

# terminal 1: local backend Worker (http://localhost:8787)
cd .\backend\
npx wrangler dev

# terminal 2: local MCP Worker (http://localhost:8788)
cd .\integrations\studyplanner-mcp\
npx wrangler dev --port 8788

# terminal 3: local Pages gateway (http://localhost:8789)
cd .\frontend\
npm run build
npx wrangler pages dev dist --port 8789

Then check:

curl http://localhost:8789/api/ai/meta
curl http://localhost:8789/privacy
curl -X POST http://localhost:8789/mcp -H "Content-Type: application/json" -d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'

Local DB refresh:

npm run db:verify-config
npm run db:migrate:local
python backend\scripts\export_sqlite_to_d1.py --data-out backend\.tmp\d1-seed.sql
cd backend
npx wrangler d1 execute DB --local --file .tmp\d1-seed.sql

Frontend checks:

npm run test:frontend
cd frontend
npm run lint
npm run build