Yes - custom prompts are supported, and they are the normal way to use OpenMontage. The demos are examples, not the required entry point. After setup, open the repo in your AI coding assistant and give it a production request directly, for example: (github.com)

Make a 60-second animated explainer about how neural networks learn

The key detail is that OpenMontage is pipeline-driven rather than demo-script-driven. For any video request, the agent should select a pipeline from pipeline_defs/, read pipeline_defs/<pipeline>.yaml, then execute each stage using the matching director skill under skills/pipelines/<pipeline>/<stage>-director.md. (github.com)

For “injecting custom prompts,” the right place depends on how deep you want to customize:

• Different topic / creative brief: no code change; just change the user prompt.
• Different script, scene, or asset-prompt behavior: edit the relevant director skill in skills/pipelines/....
• Different end-to-end workflow: add or modify a manifest in pipeline_defs/ and add matching stage skills in skills/pipelines/<your-pipeline>/. (github.com)
• Provider-specific prompt tuning: follow the tool’s agent_skills references in .agents/skills/ before generating assets. (github.com)

If my answer solved your problem, you can click answered the question. I'm really here to help, and along the way I'm also collecting Galaxy Brain badges haha 😆