Generate AI video

It is the most expensive asset there is, so the price per second is on GET /api/tools before you call. Output carries a SynthID watermark and needs an AI-generated label wherever the platform asks for one. The clip comes back as a media key ready to edit, caption or post.

Ask in plain words#

• "generate a 5 second clip of rain on a window at night"
• "AI video generation API for short-form content"
• "make b-roll of a city at dusk with sound"

From an agent host (MCP)#

Server: https://shapelessai.com/mcp (add it to your host).

• tools_run with { "name": "generate_footage", "input": { ... } }: the input is the REST body below; tools_list prints its JSON Schema.

From code (REST)#

Bearer API key with the scope each call names (auth, OpenAPI).

• POST /api/videos/generate (write scope): Generate AI footage (tool generate_footage): {name, prompt, duration_seconds?, aspect?, resolution?, with_audio?, budgetUsd?}.

From a terminal (CLI)#

• shapeless tools run generate_footage --input '<json>' (tools show generate_footage prints the schema)

Example#

curl -s https://shapelessai.com/api/videos/generate -H "Authorization: Bearer $SHAPELESS_API_KEY" -H "Content-Type: application/json" -d '{"name":"rain window","prompt":"Slow push-in on rain hitting a window at night, neon bleeding through."}'

Refusals#

Where else#

• The api docs (markdown)