One video project for people and agents
A shared video-composition substrate: edit locally in the desktop app or author the same project through MCP, REST, CLI, and x402.
CueFrame is one video composition shared by a person and an agent. Open the project in the free desktop editor to work from the transcript and timeline, or have an agent connect over MCP, REST, the CLI, or x402 to author the same clips, reframing, captions, and graphics as data. Either way, CueFrame renders production-quality video in any aspect ratio.
It is the reliable, expressive hands — not the brain. You direct; it does the legwork.
See the shared project render
Motion graphics without a flattened template. The amber title is a reusable WebGL component authored through MCP, then composited and rendered from the same project a person can open in the desktop editor. Open the template.
Subject-aware titles with real audio. The Yosemite Peregrines template places
one wide PHENOMENAL title behind the ranger while keeping its identifying
edges readable. Raised word-timed captions, the cinematic grade, original
speech, and current CueFrame logo lockup remain independent layers.
Open the template.
The loop
Every job is the same shape:
Ingest
Pull footage from a public URL or generate it from a prompt. CueFrame transcribes and detects faces so the edit lines up with the content.
Compose
Author the edit as data (clips, reframe, captions) — or hand over the intent and CueFrame drafts several cuts, scores each one, and keeps the winner. Dry-run validation is free.
Render
Render to 9:16, 16:9, 1:1, or 4:5 — baked from a single composition, never re-edited per platform.
What makes it different
- Content-aware framing — face and active-speaker tracking reframe the shot across the whole cut, not a static crop.
- One compose, every aspect ratio — output to every format from one composition.
- Scored drafts — the Director is CueFrame's authoring engine: given your intent, it drafts an ensemble of candidate edits and a server-side judge scores each one, so you get the best cut back — before you spend on a render.
- Agent-native — a curated MCP surface, async-first with webhooks, every op declarative and validateable before you spend on a render.
- One reviewable project — the editor, MCP, REST, CLI, and x402 operate on the same composition, so a human can inspect or continue an agent's work.
Next steps
Templates
Start from a finished film: one document, its looks and brand kits, applied to your footage from the desktop, an agent, the CLI or the API.
Setup
Point your agent at CueFrame over MCP, the CLI, or REST — and authenticate.
Quickstart
Intent to rendered video in three steps.
Components
The built-in catalog — every entry shown as a real render.
API reference
The full REST surface, generated from the OpenAPI spec.
Roadmap
Public priorities for the shared human-and-agent editing experience.