Skip to content
Prolifik
Video creation guide

Generate a Branded Video Batch Through MCP

Prepare exact copy for a five-video MCP batch, submit it with Prolifik and reconcile every returned piece, including failed renders and revisions.

Published 2026-10-26Facts checked 15 October 2026
An open source folder beside five distinct blue image frames arranged in a navy tray.
A batch needs a clear match between each brief and its returned file.

A good batch video request defines each piece before sending the job. “Make five videos about our app” leaves the assistant to invent the differences, and five similar hooks can turn into five versions of the same post.

Prolifik's MCP content tool can queue multiple rendered pieces under one job. Use that capability to share a stable brand and format while preserving a distinct purpose for every video. Keep the requested count, exact copy and returned piece identifiers aligned throughout the process.

The example here is a fictional five-part Text Story batch for a reservation app. It shows request construction and result handling, not a measured production test or a promise about rendering speed.

Plan five messages that do different work

The fictional app, Quiet Table, helps small restaurants manage reservation notes. The batch focuses on operational habits, without claiming that the app automatically solves them.

Piece Purpose Exact sample copy
Arrival Identify a missing detail “A booking time is not an arrival plan. Tell guests where to check in.”
Group size Make a confirmation useful “Confirm the number of seats before moving the tables.”
Access Encourage a practical question “Ask about access needs while there is still time to prepare.”
Change Clarify a handoff “When a booking changes, update the note the next shift will read.”
Close Make the lesson actionable “Before service, read the notes that could change how you welcome a guest.”

Each piece has one job. They share a voice and audience, but a viewer does not have to watch all five to understand one. This distinction matters more than varying the opening adjective.

For the creative planning behind a batch, see the batch content workflow. The technical request should preserve that plan rather than compress it into a vague topic field.

Keep the shared fields consistent

The current create_content_job interface accepts one brand and one format per job. Its count describes content pieces, not scenes or slides. If you want three Text Stories and two carousels, use separate format-specific requests instead of inventing a mixed-format array.

For the Quiet Table example, all five pieces use kinetic-type, the tool's Text Story format. They can share the saved logo choice and a restrained background treatment. The approved words differ per piece.

If the copy must come from a source page, first retrieve the selected content source and check that its analysis is ready. A source identifier should refer to actual content in the account. Do not treat a remembered title as a verified source or attach an unrelated page simply to fill an optional field.

Align each copy entry with one piece

Prolifik accepts customTexts with one entry per requested post. For this format, each entry contains a text string. Carousel entries use slides. A null entry leaves that piece's copy to the generator, so use it only when that variation is intentional.

For our five exact texts, the assistant should submit count five and five non-null text entries in the intended order. It should not send one long string containing all five messages and assume the worker will split them correctly.

A natural-language request can make the contract explicit:

Create five Text Story pieces for the selected Quiet Table brand. Use the five approved texts as five separate customTexts entries, in the listed order. Use download delivery and no music. Keep the job ID. Return a piece-by-piece result that pairs each script with its media and render status.

This request is an illustrative template. Replace the fictional brand with the real brand returned by the account and confirm the actual tool payload before creation.

Check capacity before choosing the batch size

The documented content tool supports up to 50 pieces in a job, but that maximum is not a recommendation. The account must have enough renders remaining for the requested work. Check get_usage and keep room for necessary corrections.

Start a new format with a small sample if its visual treatment has not been inspected. Once the format and brand work together, larger batches become easier to review because the remaining decisions concern the individual messages.

MCP does not bypass plan limits. It also does not make a long rendering job synchronous. The assistant receives a job and must retrieve it later, whether it asked for one piece or several.

Reconcile results at piece level

Use get_content_job to retrieve the job and its pieces. Compare the requested set with the returned scripts. Do not rely only on array position or a total count when matching outputs; keep each returned piece identifier with its text and review decision.

A partially failed job needs a precise handoff: which messages produced usable files, which failed and what the failure reason says. Recreating the whole batch can duplicate the successful work. If a failed render should use the same script, the retry-render operation is different from regeneration, which replaces a failed piece with new copy and library assets.

For a deliberately changed message, update the relevant piece and review its new render. Keep accepted pieces intact unless the shared creative direction itself needs revision.

Review a batch as individual posts

Open every returned video at phone size. In this example, the access-needs message deserves a calm background; frantic restaurant footage could work against it. The shift-handoff message must leave enough reading time for its longer sentence. A shared format does not remove these differences.

Record a simple result for each piece: accept, revise specific copy, replace unsuitable visual treatment or hold. The Text Story format page gives context for the output being reviewed. Do not call the batch finished because its job stopped processing.

The Prolifik MCP guide documents the current creation, retrieval and editing tools. A completed batch should leave you with the intended distinct videos and a usable mapping from brief to file. Scheduling can then use the accepted piece identifiers without accidentally selecting an earlier draft.

Build a batch with distinct messages

Use the current content-job tools to keep approved copy, returned pieces and review decisions aligned.

Read the MCP content tools
Sources and verification

Product and primary-source checks: 15 October 2026. Worked examples are illustrative unless a real source is identified.