Documentation for AI assistants and tools

Use the same Zvid documentation in your browser, an AI conversation, an API client or a local validator. The HTML reference and Markdown exports contain the same technical material. You do not need to execute page JavaScript to read API fields and examples.

Start with the task

Task	Read first
Render a first video	Quick Start, then validate and estimate
Render a still image	Rendering images
Write project JSON	Project structure, then the relevant element reference
Connect ChatGPT or another assistant	AI assistants
Personalize one design	Templates and dynamic content
Operate a production integration	Render lifecycle, errors and retries, webhooks

Downloadable formats

Resource	What it contains
LLM orientation	Key facts and a short route into the documentation
Complete Markdown index	Every documentation page with its purpose and links
JSON index	Page metadata for indexing and tool integration
Full documentation export	All pages with source boundaries; useful for offline ingestion
OpenAPI specification	Public operations, authentication, requests, responses and examples
Authoring project schema	Project input, including supported template-authoring constructs
Resolved project schema	The concrete project after variable/iteration resolution
Render request schema	The outer request: either payload or template, plus supported options

Each documentation page has a Markdown link. For example, the Quick Start Markdown contains the guide's text and code without navigation or interactive controls. Start with individual pages instead of sending the entire export when your task needs only a few topics.

Which contract should I use?

The request envelope is what you POST to the API. Its payload is a project, and the project's visuals contain elements. A code block labelled “element fragment” is not a complete API request.

For templates, authored input can contain placeholders, variables, conditions and iteration. These resolve before final project validation. A resolved-project schema should not be used to reject an unresolved template solely because it contains authoring constructs.

Static schemas help catch structural errors. They do not replace account-specific limits, cross-field rules, media access checks or actual rendering. Use free server validation for the request you intend to submit. GET /api/render/schema/api-key supplies current, account-aware schema guidance.

Facts to preserve when generating code

• The API origin is https://api.zvid.io; documented operation paths include /api/.
• REST API keys are created at app.zvid.io/api-keys and sent as x-api-key.
• A render request contains exactly one of payload or template.
• Rendering spends credits and is asynchronous. Submission returns jobId; job lookup reports state.
• Free validation and template preview do not queue render jobs.
• A signed registered webhook and a one-off webhookUrl callback have different verification properties.
• Use the response shape documented for the specific operation; list/get/create endpoints do not all share the same wrapper.

Freshness and citations

Exports are generated with the website and carry source information. Cite a page's canonical HTML URL when answering a user, and use its Markdown equivalent when clean source text is useful. If a live authenticated validation result differs from a downloaded schema, keep the returned field errors and check the current reference before submitting a paid render.

The LLM index is a discovery aid. It does not grant access to an account or guarantee that a particular assistant will retrieve or cite a page.