The fastest way to find answers is the official documentation:
| Channel | Use for |
|---|---|
| GitHub Discussions | Questions, ideas, show-and-tell |
| GitHub Issues | Bug reports and feature requests |
| Discord (link in README) | Real-time chat with maintainers and the community |
squish: command not found
Make sure the pip scripts directory is in your PATH. For a virtual environment, activate it first (source .venv/bin/activate). For Homebrew, run brew link squish.
Model download is slow or interrupted
Run squish pull <model> again — downloads resume automatically.
ImportError: No module named 'mlx'
MLX only runs on Apple Silicon. Squish requires macOS 13+ on M1–M5 hardware.
Server not responding
Check that the server started with squish run <model> before calling the API, or use squish daemon status.
Out of memory during inference
Try a smaller model (e.g. llama3.1:3b instead of llama3.1:8b) or close other applications to free unified memory.
Please do not open a public GitHub issue for security vulnerabilities.
Report them privately via GitHub Security Advisories.
For anything not covered above, open a GitHub Discussion or a GitHub Issue.