Local AI Cat module
omlx sidecar (Python MLX oracle)
Current shipping, testing, and product context for this part of Local AI Cat.
Status
Where it stands
- Platforms
- macOS
- Release channels
- Alpha
- Distribution
- Not currently distributed
- AlphamacOS
Description
What it does
macOS / Alpha only. Bridge track from docs/planning/done/omlx-sidecar-plan.md: FeatureOMLXSidecar's SidecarProcess spawns/health-polls/tears down the omlx server, SidecarBackend adapts its /v1/chat/completions SSE (incl. reasoning_content + streamed tool_calls) into LocalAITextGenerationEvents, and SidecarRouting sends big/coding downloaded MLX models to the sidecar (H1: unloads the in-process runtime first). OMLXSidecarRegistration owns ChatBackendRegistry registration, while OMLXSidecarTerminationParticipant owns process teardown; no app-side sidecar startup or parallel lifecycle path remains. Engine policy via localAI.engineSelectionMode (auto/forceNative/forceSidecar) in Advanced Settings lets us A/B the native runtime vs omlx through the same app surfaces. Deferred: S4 runtime bundling (currently needs a dev-machine omlx venv).
Headless Local API
Drive it over the loopback API
Logging / observability: provisional grade L0 — structured=0 print=0 sentry=0.
- discover
- not applicable
- execute
- present
- observe
- not applicable
- configure
- not applicable
- cancel/stop
- not applicable
- reset
- not applicable
/v1/local/app/action/v1/local/app/artifacts/{id}/v1/local/app/navigate/v1/local/app/screenshot/v1/local/app/state/v1/local/app/surfaces/v1/local/app/transactions/{id}/v1/local/artifacts/v1/local/artifacts/{artifact_id}/v1/local/artifacts/{artifact_id}/content
Media
See it in action
Backlog / improvements
What comes next
No improvements are listed for this module yet.
History
How it got here
No change history has been published for this module yet.

