Stage definitions
A YAML manifest describes production steps, the tools involved and the conditions to check before moving on.
OpenMontage gives a coding agent production plans and tools for moving a video from brief to rendered cut, with review points for footage, audio and captions.
Use OpenMontage’s staged production workflow to plan a short demo around the actions your viewer needs to see. Keep the footage, caption plan and final checks traceable.
List the source recording, essential actions, duration, audience and output aspect ratio.
Copy the request into a new Floatboat conversation and provide the required material.
Download FloatboatPreview and copy the request. It is not sent automatically from this page.
Reference: OpenMontage https://github.com/calesthio/OpenMontage Read the original documentation and check the current environment and dependencies. Propose any new installation or tool connection before proceeding. My task: Review the OpenMontage repository and help plan a 45-second demo from my existing screen recording. Keep onboarding, search and the export confirmation visible. Propose a timestamped edit, flag caption overlaps and list checks for the 9:16 export. Ask before sending footage to an external service. Expected deliverable: An edit brief with timestamped scene choices, caption notes and a pre-delivery checklist.
A product demo needs a script, screen recordings, captions and a final cut. OpenMontage breaks that work into stages with YAML pipeline definitions and separate instructions for the coding agent. The agent reads the plan, calls the relevant tools and records decisions and progress so later steps can use the same production context.
Different projects take different paths. An explainer may use generated stills and a Remotion composition, while a documentary montage can retrieve real motion clips from open archives and use them to assemble a timeline. The available providers, credentials and asset rights determine which path you can take.
For a Floatboat workflow, use the project as a reference for sequencing a brief, asset approval and cut review. Keep the intended delivery format visible from the beginning, then inspect sampled frames, captions and audio against that brief after rendering.
The pipeline leaves a record of which tools and assets were selected and when a result was checked. Use it to review choices that affect the finished cut, such as the footage source and composition engine.
A YAML manifest describes production steps, the tools involved and the conditions to check before moving on.
The agent can use source footage, local tools or configured cloud services, with selection rules that account for fit and cost.
The project uses tools such as ffprobe, sampled frames and audio analysis to flag technical issues for review.
Pick the output first, from a real-footage montage to a captioned demo. Then choose the stages and inspection points that protect what the audience needs to see or hear.
Start with a screen recording and a list of actions the viewer must see. Use the pipeline to draft the cut and captions, then verify that captions stay clear of controls and the edit preserves each essential step.
Specify real footage and a theme before sourcing. Track the origin and usage conditions of each candidate, then review whether the cut combines actual moving clips at the intended pace.
Approve the script and scene plan before producing visual assets. Render a short sample and listen for awkward pronunciations, then verify that scene changes still line up with narration after any revisions.
Retain the source transcript and use it to review the translated script before producing a new voice track. Check subtitle wrapping and time alignment in the localized render, especially around rapid speech or labels.
The upstream quick start runs inside a coding-agent workspace. Prepare the runtime and approved inputs before selecting a pipeline, and enable provider credentials only for services your production needs.
The source quick start lists Python 3.10+, Node.js 18+, FFmpeg and an AI coding assistant.
Gather approved footage or source links, and specify duration, aspect ratio, audience and the expected output format.
Some paths use local tools and open sources. Configured cloud models and some stock services require separate credentials and may incur charges.
Compare the workflows, operating requirements, and source licenses of these projects.