Prompt Compass is a lightweight routing and safety decision layer for AI
applications. It evaluates incoming prompts before they reach a configured
response model and can return a local-routing decision, a cloud-routing
decision, or selected safety-block decisions. Developers choose their response
models and endpoints. Prompt Compass does not generate the final response.
What Prompt Compass does
Prompt Compass helps applications decide where a prompt should go before a
response is generated:
Local — use a configured local response model when that fits the task.
Cloud — use a configured cloud response model when that is appropriate.
Block — stop selected prompts when a safety decision requires it.
The extension can dispatch prompts to the local or cloud endpoints you
configure. You remain responsible for choosing and evaluating those models.
What Prompt Compass does not do
It does not generate the final response.
It is not an LLM or model provider.
It is not an inference provider.
It does not universally replace cloud models.
It does not guarantee savings; any savings depend on the workload, routing
outcomes, model quality, and deployment economics.
It is not by itself a complete security, privacy, or compliance system.
