Requirements
- Node.js 16 or later (16, 18, 20, 22, and 24 are all supported)
- npm 8 or later
- Ollama installed on your system to pull and run models
Install
npm install -g llm-checker
After installation completes, npm runs a postinstall script that prints:LLM Checker installed. Run: llm-checker hw-detect
This is expected — it is just a reminder of the first command to run.
Verify the installation
You should see the current version number (e.g. 3.5.6).
Optional: sql.js for database features
Three commands — sync, search, and smart-recommend — require an optional SQLite dependency. The core tool works without it.
Once installed, these commands become available:
| Command | Description |
|---|
sync | Download the latest model catalog from the Ollama registry |
search <query> | Search models with filters and intelligent scoring |
smart-recommend | Advanced recommendations using the full scoring engine |
GitHub Packages (legacy)
The GitHub Packages mirror (@pavelevich/llm-checker) may lag behind the npm release. Use the npm package for the latest version.
If you specifically need to install from GitHub Packages:
Configure the registry and token
You need a GitHub Personal Access Token (PAT) with read:packages scope.echo "@pavelevich:registry=https://npm.pkg.github.com" >> ~/.npmrc
echo "//npm.pkg.github.com/:_authToken=${GITHUB_TOKEN}" >> ~/.npmrc
Install the scoped package
npm install -g @pavelevich/llm-checker@latest
Switch back to npm if the version looks old
If the installed version is older than expected, migrate to the npm package:npm uninstall -g @pavelevich/llm-checker
npm install -g llm-checker@latest
hash -r
llm-checker --version