Jimeng Video Query Tasks

Every cloud video node in this pack submits a task to ByteDance's Ark API, and tasks have a life of their own after you hit "Queue." This node is how you inspect that life: it lists your video task history, filters it by status, model, or ID, and hands you the results as JSON. If you've been using non_blocking mode - where clips keep rendering in the background while your queue moves on - this is the node that tells you what actually finished.

It's a utility node, and it knows it. No generation here, no images to fiddle with; it's a query tool, closer to the Jimeng Quota Settings node's practical cousin than to a generation node.

How it works

The node wraps the Ark SDK's task-list endpoint. You give it filters, it calls the API (with page_num/page_size pagination under the hood) and returns the matching task list. The task_ids field is the interesting one: paste one or more task IDs separated by newlines and it queries exactly those tasks - the canonical way to check on non_blocking jobs you submitted earlier. If you leave it empty, you get a filtered page of history instead.

The inputs that matter

• status - filter by lifecycle state: all, succeeded, failed, running, queued, cancelled, expired. "Queued" and "expired" are worth knowing about: flex-tier jobs can sit queued, and tasks expire if not picked up.
• model_version - narrow to a specific Seedance model, or all.
• service_tier - default or flex (matches the enable_offline_inference toggle on the generation nodes).
• task_ids - newline-separated IDs for targeted lookups.
• page_num / page_size - pagination; default page size is 10.
• seed - present but you'll rarely touch it; used for stable ordering of results.

Outputs

• task_list_json - the matching tasks as a JSON string. Pipe it through a Text node or just read it in the console.
• total_tasks - the count, as an integer you can wire into other nodes if you're building something programmatic.

Installing and using it

Install is the pack-standard route: ComfyUI Manager (search "ComfyUI Jimeng API") or git clone https://github.com/fkxianzhou/ComfyUI-Jimeng-API into custom_nodes, restart, and let the SDK self-install. No models, no extra setup. You need a client from the Jimeng API Client node, same as every other node in the pack.

Two practical notes. First, this node queries the API's task store, so it sees everything your key has submitted - including jobs from the non-blocking runs - not just the current workflow. Second, it only covers video tasks; image and visual-understanding jobs aren't in this list. If you're debugging why a background render never appeared, this is where you'll find the answer: failed with a message beats staring at an empty output folder.