create-video-implementation-workflow

create-video-implementation-workflow

Purpose

Create the build instructions that a coding agent can execute to produce the final video without making production decisions.

Use after concept/script approval and before generating HTML, voiceover, music, or final edits.

Inputs

• Approved concept brief and script.
• Brand aesthetic and raw-materials manifest.
• Required tools/providers such as Hyperframes, FFmpeg, goose-graphics, ElevenLabs, or local music generation.
• Output folder.

Workflow

1. Create a scene table with timing, visual description, caption, VO line, source assets, and generation method.
2. Assign tools per scene: HTML/Hyperframes, goose-graphics PNG, website asset, generated frame, voiceover, music, or FFmpeg assembly.
3. Define folder layout for raw materials, generated assets, audio, clips, edits, review frames, and manifests.
4. Specify exact verification commands: ffprobe dimensions/duration, sampled frames, audio duration, and visual review.
5. Include failure handling for missing credentials, inaccessible assets, provider failures, and render collisions.
6. Keep all steps programmatic; do not include manual editor tools such as Premiere, CapCut, or DaVinci.

Output

• workflow.md with scene-by-scene instructions and commands/pseudocode.
• Tool/provider checklist.
• QA checklist and acceptance criteria.
• manifest.json and verification.md for test or production runs.

Quality Checks

• Another agent can execute the workflow without choosing tools, timings, file names, or scene order.
• Each scene names its source material or generation method.
• Final export requirements are explicit.
• Human approval gates are named where brand or generated media review is required.

Failure Modes

• Concept brief is not approved or lacks timings.
• Workflow leaves asset sourcing, tool choice, or export settings ambiguous.
• Provider usage is incompatible with available credentials.
• QA steps do not verify playable output.