ImStudium

Im Studium — alles griffbereit. Full study system for Stud.IP students in Germany — extract course materials, sync to ~/Studium, plan Klausuren, and use with any LLM via MCP.

Onboard (first time)

One line (no Bun / no global install):

curl -fsSL https://raw.githubusercontent.com/paddy-shrp/imstudium/main/scripts/install.sh | bash

Or:

npx -y @imstudium/cli onboard -y -o ~/Studium

First login (opens your normal browser, no extra tools):

imstudium login

One command: login → sync → wire Cursor, Claude Code, Claude Desktop/Cowork, Codex, Windsurf, OpenCode (only detected harnesses). Idempotent — re-running skips unchanged skills and MCP.

Upgrading from DigiCampus? Migration is automatic — imstudium setup migrate -y copies your session, config, and rewires agents.

Daily use:

imstudium sync --quiet --extract

Agent setup only (once, or after upgrade):

imstudium setup all -y          # skips unchanged
imstudium setup status          # what’s configured

What you get

~/Studium/
├── llms.txt           # agent entry point — paths + search guidance
├── studium.json       # machine-readable course map (schema v2)
├── STUDIUM.md         # semester overview
├── klausuren.md       # exam timeline with days-remaining
├── calendar.ics       # full schedule
└── SS2026/            # active semester (slug from Stud.IP)
    └── Analysis-II/
        ├── COURSE.md
        ├── search/
        │   ├── materials.json   # file list + cite refs
        │   └── corpus.jsonl     # page index — grep this with your agent tools
        ├── files/       # synced PDFs, slides, sheets
        └── extracted/   # per file: content.md + pages/*.png
            └── Vorlesung-03-pdf/
                ├── content.md
                ├── pages/page-001.png ...
                └── manifest.json

Sync defaults to the current semester. Other terms: imstudium sync --semester "WS 2025/26".

Quick start

# First time (recommended)
npx -y @imstudium/cli onboard -y -o ~/Studium

# Or: workspace first, agents second (less churn on re-sync)
imstudium init -o ~/Studium
imstudium onboard --agents-only -y

# Daily
imstudium sync --quiet --extract
imstudium --json studium

CLI commands

Command	Description
onboard	First-time: login + sync + agents (idempotent)
init	Login + sync workspace only
setup status	Detected agents + manifest
setup all	Wire agents (skips unchanged)
login / logout / status	Auth lifecycle
studium	Overview (--json for studium.json)
klausuren	Exam timeline
courses list	Live course list
sync	Sync files + rebuild workspace (--extract to also extract; --semester for other terms)
extract	Text + page images from all synced materials
index	Rebuild page index (corpus.jsonl) for agent grep / semantic search
view	Build viewer.html for Playwright / browser agents (fast)
doctor	Check extraction tools (poppler, LibreOffice)
update	Upgrade CLI + refresh agent skills/MCP
watch --interval 1h	Continuous sync
discover	List JSON:API routes

Global flags: --json, --quiet, --instance <url>

MCP (all agent harnesses)

imstudium setup mcp    # merges into each harness MCP config (see AGENTS.md)

Or copy mcp.json. See AGENTS.md for per-harness paths.

Tools: studium_overview, studium_klausuren, studium_course_brief, studium_find_materials, studium_read_material, imstudium_sync, imstudium_sync_dry_run, imstudium_list_courses, imstudium_whoami

Content search: grep or semantic-search <course.path>/search/corpus.jsonl with your agent — studium_overview returns searchGuidance. No custom search MCP tool.

Extraction — agents that SEE your slides

Every synced file is extractable to text and rendered page images:

brew install poppler                  # PDF text + slide PNGs
brew install --cask libreoffice      # optional: PPTX/DOCX/XLSX rendering
brew install tesseract tesseract-lang # optional: OCR for scans & photos
imstudium doctor                     # verify tools

imstudium extract                    # everything
imstudium extract --course Analysis  # one course
imstudium sync --extract             # sync + extract in one step

Input	Text	Page images	Tool
PDF	✅ per page	✅ PNG per page	poppler
Scanned PDF (no text layer)	✅ OCR per page	✅ PNG per page	poppler + tesseract
PPTX / DOCX / XLSX / ODP...	✅ per slide	✅ via LibreOffice → PDF	libreoffice + poppler
PPTX / DOCX (no LibreOffice)	✅ native XML	—	built-in
Images / photos (PNG/JPG/...)	✅ OCR if tesseract	✅ copied to pages/	built-in (+ tesseract)
Text / code / CSV / notebooks	✅ passthrough	—	built-in

All tools are free and local — no API keys, no credits, no cloud. Everything degrades gracefully: missing tools just mean fewer output layers, never a crash (imstudium doctor shows what's active).

MCP agents use studium_read_material — returns requiredAction, pageImages, and viewerFileUrl. Read every PNG or open the viewer in Playwright/browser before explaining slides.

Agent skills

Bundled in the CLI package (like Firecrawl). Install with:

imstudium setup all -y

Skill	Purpose
imstudium-cli	Terminal commands
imstudium-studium	Course overview
imstudium-klausuren	Exam timeline
imstudium-sync	Material refresh
imstudium-visual	Read PNGs / Playwright — mandatory for slides
@imstudium/mcp	MCP tool picker
klausur-prep	Exam prep workflow
semester-refresh	Daily briefing workflow

See packages/cli/skills/ · AGENTS.md · CLAUDE.md

Packages

Package	Description
@imstudium/sdk	JSON:API client, auth, sync, Studium builder
@imstudium/cli	CLI
@imstudium/mcp	MCP server

Auth

Stud.IP (e.g. DigiCampus at Uni Augsburg) uses Shibboleth + Microsoft MFA.

1. Clean login: imstudium login --oauth — one-click app login once a university OAuth client is registered (docs/OAUTH.md).
2. Temporary fallback: imstudium login — opens a local helper page and your normal browser. This currently requires a session value because a CLI cannot securely read a normal browser session by itself.
3. Cookie import: imstudium session import --cookie <Seminar_Session>

Instance

Default: digicampus.uni-augsburg.de (Stud.IP 5.4.11)

Other Stud.IP universities:

imstudium config set base-url https://studip.example.edu/jsonapi.php/v1

Legal

For personal study use only. Respect university terms of service and course material licenses.

License

MIT — Copyright (c) 2026 Patrick Schröppel