vidhook-movie

vidhook: authoring Movie JSON

vidhook is a rendering primitive: you give it a Movie JSON (scenes, images/video/text/audio, transitions) and it renders an MP4. It does not generate or host assets. Every src is an http(s) URL you provide — host them yourself (your own CDN/storage) or get URLs from a generation MCP, then reference them. The example URLs here (https://cdn.example.com/...) are placeholders; replace them with real, reachable assets.

This skill is the brain (how to write correct Movie JSON). The vidhook-mcp server is the hands (validate, render, get_status tools that call the vidhook API). Use them together.

Workflow — always validate first

1. Author a Movie JSON (see patterns and examples/ below).
2. validate the Movie. This consumes no credits and starts no render. It returns either field-level errors or estimatedCredits. Fix errors and re-validate until it is valid and the cost is acceptable. estimatedCredits is exactly what render will reserve.
3. render a draft using a vh_test_ key (free tier → watermarked output) to preview the actual video. It returns renderId + bucketName.
4. get_status with that renderId + bucketName, polling until done is true. When done and not fatalErrorEncountered, outputFile is the result video URL.
5. Finalize with a vh_live_ key (paid → clean, no watermark) once the draft looks right.

Never skip step 2. Validating first is free and catches every schema mistake before you spend credits or wait on a render.

API key axes (set via environment, never tool arguments)

Two independent axes, both configured on the MCP server via environment variables:

• Key type → watermark & billing: vh_test_… = free tier, watermarked (drafts); vh_live_… = paid, clean (finals).
• Base URL → environment: VIDHOOK_API_BASE_URL selects which environment you talk to. It does not change watermarking. A vh_test_ key always watermarks, regardless of base URL.