| name | run-anime-upscaling |
| description | Launch, verify, and stop the Anime Upscaling app (Go API + Next.js dashboard, Docker Compose). Use when asked to run, start, launch, boot, or smoke-test this project locally, or to confirm a change works in the real app. |
Run Anime Upscaling
This project is a self-hosted Go API + Next.js dashboard for anime video
processing, run via Docker Compose. The canonical, always-up-to-date runbook is
AGENTS.md; this skill is the short path to launch it.
Launch (CPU, prebuilt images — fastest)
make quickstart
What it does: generates AUTH_SECRET + a random AUTH_PASSWORD into .env,
creates the data/* folders, starts the stack from prebuilt Docker Hub images
(no build), and prints the login password. Idempotent — safe to re-run.
For a local build / GPU run instead:
make run
make run-gpu
Verify it's up
docker compose ps — anime-upscaling-app and anime-upscaling-api are up.
- Open http://localhost:4750 and log in with the
AUTH_PASSWORD from .env
(grep '^AUTH_PASSWORD=' .env).
- Health probe: the API exposes
GET /api/health/gpu (reports healthy on
CPU-only too).
- To exercise a job: drop a video into
data/input, queue an optimize job
from the UI, and watch the per-job log stream.
Logs / stop
make logs
make stop
Must-know constraints
- The Go API has no auth and must stay internal — only the app port
4750
is published; never expose API port 4751.
- Jobs are in-memory (lost on API restart); only pipeline definitions
persist.
- GPU processing needs the NVIDIA Container Toolkit on the host; without it, use
the CPU path (
make quickstart).
See AGENTS.md for the full runbook, testing commands, and
contribution rules.