Problem Statement
A core selling point is running reviews on your own infrastructure against a local model (Ollama) so no code leaves the network — but there is no guide for an air-gapped/offline deployment, and it is unclear which outbound calls (telemetry, GitHub, provider) happen and how to disable the optional ones.
Proposed Solution
Document an air-gapped/offline deployment: required vs optional outbound connections, how to point at a local Ollama model, how to disable or self-host OTLP telemetry, and how the dashboard behaves without external calls. Add a telemetry opt-out switch if one does not already exist.
Alternatives Considered
- Leave it to users to infer from config — undercuts the privacy niche the project targets.
Priority
Nice to have
Code of Conduct
Problem Statement
A core selling point is running reviews on your own infrastructure against a local model (Ollama) so no code leaves the network — but there is no guide for an air-gapped/offline deployment, and it is unclear which outbound calls (telemetry, GitHub, provider) happen and how to disable the optional ones.
Proposed Solution
Document an air-gapped/offline deployment: required vs optional outbound connections, how to point at a local Ollama model, how to disable or self-host OTLP telemetry, and how the dashboard behaves without external calls. Add a telemetry opt-out switch if one does not already exist.
Alternatives Considered
Priority
Nice to have
Code of Conduct