get started
Install, pull, run.
Three steps from zero to a guarded LLM call: install the binaries, pull the guard models, run a check. Works offline once the models are cached.
-
01 Install the binaries
Go 1.22+. Each binary is independent; install only what you need.
-
02 Pull the guard models
On Linux and Windows, reed pulls both models via Ollama in one step. On macOS, omlx loads them from the local MLX directory — no pull needed.
-
03 Run a guarded search
The first call wires up the backend, confirms the models respond, and writes guarded markdown to stdout.
$ reed search "constructed wetlands filtering" [leakage] pass [brave] 12 urls [shield] pass [canary] pass ✓ 3.4kb markdown → stdout