chrissoria/catllm-desktop

CatLLM Desktop

A native macOS app for classifying open-ended survey responses, PDFs, and images using Large Language Models. Wraps the cat-llm Python package in a self-contained bundle — no Python install required.

Install (first time)

1. Download the latest DMG from Files and versions:
   • Apple Silicon (M1/M2/M3/M4): CatLLM-X.Y.Z-arm64.dmg
   • Intel Macs: CatLLM-X.Y.Z-x86_64.dmg
2. Open the DMG and drag CatLLM to your Applications folder.
3. First launch only: macOS will show "Apple cannot check it for malicious software." This is expected for unsigned indie apps.
   • Right-click (or Control-click) CatLLM in Applications → Open → click Open in the dialog.
   • After that one time, it launches like any other app from Spotlight, Launchpad, or the Dock.

Verify your download (optional)

Each DMG has a matching .sha256 file. Verify with:

shasum -a 256 -c CatLLM-X.Y.Z-arm64.dmg.sha256

What you can do

Function	Purpose
Classify	Assign categories (manual or auto-extracted) to text / PDF / images
Extract	Discover categories from your data
Explore	Saturation analysis for category discovery
Summarize	Generate concise summaries

Across 9 LLM providers (OpenAI, Anthropic, Google, Mistral, Perplexity, xAI, HuggingFace, Ollama, and more) and 7 domain packs (General, Survey, Social Media, Academic, Policy, Web, Cognitive).

Your data stays local

The app runs entirely on your machine. Only the text you classify is sent to your chosen LLM provider's API. Do not upload sensitive, confidential, or personally identifiable information when using cloud providers. For fully local inference, use the bundled Ollama integration.

Getting an API key

You'll need at least one provider key. Most users start with:

• OpenAI — platform.openai.com/api-keys
• Anthropic — console.anthropic.com/settings/keys
• Google AI Studio — aistudio.google.com/apikey

Paste keys in the Cloud API Keys panel in the app sidebar. They're stored locally on disk only.

Source code

• Python package: github.com/chrissoria/cat-llm
• Desktop wrapper: same repo, app/ directory

Citation

Soria, C. (2025). CatLLM: A Python package for LLM-based text classification.
DOI: 10.5281/zenodo.15532316

License

AGPL-3.0.