Question 1

Why scan local LLM configs for privacy?

Accepted Answer

Local LLM tools like Ollama, LM Studio, and Jan may phone home with telemetry, embed API keys in config files, or expose local paths. Scanning before sharing configs prevents accidental data leakage.

Question 2

What configs are supported?

Accepted Answer

Ollama Modelfiles (FROM, PARAMETER, SYSTEM), LM Studio JSON configs (model paths, server settings), Jan AI settings.json, and generic JSON/YAML config files.

Question 3

Is my data sent anywhere?

Accepted Answer

No. All scanning runs in your browser with zero network requests. The live network counter on the page proves it stays at 0. No cookies, no localStorage, no analytics.

Question 4

What risks are detected?

Accepted Answer

Six categories: Outbound endpoints (URLs that phone home), telemetry flags, hardcoded API keys (OpenAI, Anthropic, AWS, HuggingFace), auto-update/callback URLs, data leakage (local paths, log files), and insecure settings (CORS *, disabled SSL, 0.0.0.0 binding).

Question 5

How to fix detected issues?

Accepted Answer

Disable telemetry flags, move API keys to environment variables, restrict CORS and host binding to localhost, and remove callback URLs you don't recognize.

Local LLM Privacy Probe