Use case
Short vertical clips on a schedule
Put clip production inside a scheduled job. A script goes in, a run id comes back, a file lands in your storage. Retries are free until the vendor bills.
The pipeline, end to end
Something decides what today's clip says. A script is written, by a person or by a model you already run. The script and a person go to Clipwright, a run id comes back, and a job polls it until a file is ready. The file lands in your storage.
None of that needs a browser open, which is the only reason any of it can be scheduled.
Why retries matter more here than anywhere else
An unattended job retries. It has to, because transient failures are normal and nobody is watching to intervene.
Billing starts when the vendor accepts the paid render of your clip, and not before. A run that breaks earlier costs nothing, and a run that breaks because of us costs nothing either, even when we have already paid the vendor.
This is what makes a retry loop safe to leave running. Without it, every automated retry is a small bet.
Cost scales with seconds, nothing else
30 credits per second of finished video. A daily clip of a given length has a knowable monthly cost, and it does not move because of which options you used.
Credits come from a pack rather than a monthly allowance, so a week where the job produced nothing does not cost anything.
The part that should not be automated
A scheduled job will happily produce a clip every day whether or not there is anything worth saying. Volume without judgement is how a feed turns into noise that nobody watches, including the algorithm.
The honest version of this page says: automate the production, keep a person on the decision. The tool cannot tell you that, so we are saying it here.
Collect and publish yourself
Files are kept for 30 days, then deleted. Fetch them in the same run that created them.
Clipwright posts nowhere. Whatever handles distribution in your stack continues to handle it.
Start
A job that posts daily without anyone reading the output is how a feed stops being worth following. Wire one scheduled run, watch a week of what it makes, and decide then whether it deserves to run unattended.
Checked on 2026-09-15.
Questions
- Can this run unattended?
- The rendering can. Submit returns a run id, polling tells you when it is done, and both terminal states are explicit. What should not run unattended is the judgement about what is worth saying, which is the part that decides whether any of it works.
- What does a daily clip cost?
- 30 credits per second of finished video. Work out your daily seconds and multiply; there is nothing else in the calculation.
- Will a scheduled job burn credits on failures?
- You are charged only once the vendor has accepted the paid render of your clip. A failure before that point costs nothing, and so does a failure caused on our side, even when we have already paid the vendor. A retrying job is not a spending risk.
- Does it post the clip for me?
- No. Clipwright renders the file and returns it. Publishing stays in your own stack, wherever you already handle it.
- How long do I have to collect the file?
- Clips are kept for 30 days after the run and then deleted, and links expire sooner. Collect the file in the same job that created it.
Render a clip and compare the files yourself.