CD · Queue Checkpoint

The worst feeling in batch generation isn't the failure - it's the failure at take 37 of 40, followed by re-running all 37 because there's no record of what finished. CDQueueCheckpoint is the node that makes that stupid tax disappear. You give it your execution plan and the list of task ids that finished (and which failed), and it computes a resumable checkpoint: exactly what's left, in the original order, so the next run can pick up where the last one died.

It's from Continuity Director, a 20-node pack for repeatable AI video production, and it's one of the reliability nodes - the answer to "how do I resume a long batch."

What you feed it

• execution_plan - a CD_EXECUTION_PLAN wire from CDExecutionPlan. This is the ground truth for the task list and its order.
• completed_ids_json - a JSON array of task ids that finished, default [].
• failed_ids_json - a JSON array of task ids that failed, default [].

How it works

The node walks the plan's waves to get the full ordered task list, then marks the completed and failed ids against it. It's strict in exactly the ways you want: a task can't be both completed and failed, and unknown task ids are rejected - typo a take id and it refuses rather than silently mischeckpointing. Overlaps between completed and failed throw a clear error listing the offending ids.

Outputs: checkpoint (a CD_CHECKPOINT wire), checkpoint_json (the full record - completed, failed, remaining, each in plan order, plus remaining_count and a hash), and remaining_count (an integer - the number of tasks still to run). That last number is your progress bar, and the record is your resume file: save the JSON, restart, feed it back with the new completed list, and you've survived the crash.

Installing it

Continuity Director is a frictionless install: no models, no API keys, zero Python dependencies (requirements.txt literally just declares that). Search "Continuity Director" in ComfyUI Manager and restart, or:

cd ComfyUI/custom_nodes
git clone https://github.com/xinjian0101/continuity-director.git ComfyUI-ContinuityDirector

Restart ComfyUI - nodes appear as CD · ..., plus a Continuity Director sidebar with an "Add starter chain" scaffold. Update with git pull.

The honest take

Two things to keep straight. First, this node computes the checkpoint - it doesn't store it or watch the queue. You own saving the checkpoint_json somewhere persistent (and feeding fresh completed ids back in on resume); that's the discipline the pack assumes you bring. Second, failed tasks are excluded from remaining, so a failed take won't magically reappear in the resume list - you'll want to pair this with CDRetryPolicy or your own retry to decide whether failed means "retry" or "give up." Get those two habits right and this is the node that turns a 40-take batch from a thing you babysit into a thing you can walk away from.