Mother AI

What it does

Authenticates requests via Authorization-header API tokens against team policy.
Persists projects and workflows.
Enqueues every chat submission as a Redis job before any execution — the queue-first invariant.
Streams job events to clients over Server-Sent Events.
Probes models and wakes Ollama on demand.
Routes non-LLM intents (rules, facts, commands).

Method	Path	Purpose
`POST`	`/v1/chat`	Submit a prompt. Returns `job_id`.
`GET`	`/v1/jobs/:id`	Current job status.
`POST`	`/v1/jobs/:id/resume`	Resume a job paused on an `ask_user` interrupt.
`GET`	`/v1/jobs/:id/stream`	SSE: replays Redis backlog, then streams the live pubsub channel until the job finishes or errors.
`GET`/`POST`	`/v1/workflows`	Workflow graph CRUD.
`GET`/`POST`	`/v1/projects`	Projects CRUD.
`GET`	`/v1/models`	Available models — local Ollama plus configured Claude / OpenAI-compatible.
`GET`/`POST`	`/v1/ollama/*`	Ollama health and wake.
`POST`	`/v1/command/intents`	Non-LLM intent routing.
`GET`	`/healthz`	Liveness.

The stream handler must replay the durable backlog before joining the pubsub stream so that a reconnecting client never misses events.

Errors include job and request correlation metadata so clients can trace any failure back to the job and the original request.