MCP App Store
by KeenEthics
Scenivia icon

Scenivia MCP Server for ChatGPT

By Scenivia

Overview

Create launch and feature videos for your SaaS product from its website. Scenivia turns a product brief into an editable motion-design film of 15 to 45 seconds for X, LinkedIn, Product Hunt or a landing page. Create a project, set the brief and brand, pick a template from the Library and write its lines, change one scene at a time, then check and export the MP4. Cards show the project, the export progress and the templates inline. Scenivia acts with the connected account, only on the projects that account can open, and starts an export only after the person agrees.

How to connect

  1. ChatGPT: Open the Scenivia listing in the ChatGPT app directory and choose Connect. ChatGPT listing Scenivia exposes 33 tools here.

Tools

add_asset_from_url

ChatGPT
Fetch a public PNG, JPEG or WebP image and add it to the project, then use its id with setAsset. The same checks as an upload apply; the same picture twice is the same asset.

apply_turn

ChatGPT
Apply a proposed turn, once. Property edits are applied at once; shot changes, composed screens, template films and film edits are handed to the Studio (state awaiting_studio) and get_turn reports their revision. Refused if the film changed since baseRevision or the turn was already applied.

check_export

ChatGPT
Translate the saved film for the renderer without rendering or charging: shot count, played length and what the export would lose, or why it cannot export.

compose_screen

ChatGPT
Experimental: prefer a real screenshot. Validate a product screen no template carries: up to 6 blocks (window, card, list, button, badge, text, image) on a 12 by 8 grid, in the brand's colours, reflowed for 1:1 and 9:16; or a source screen from rebuild_screen_from_capture, keeping the page's words. Added as the last shot, or given to an existing composed screen with scene. Refused with each issue and a hint: fix them and call again. Returns a turn to apply, which the Studio performs.

create_project

ChatGPT
Create an empty personal project. Each call creates a new one.

edit_film

ChatGPT
Validate what the Votre film panel changes for the whole film, as the person would: the brand (setBrandName, setProductName, setLogo, setBrandColours, resetShotColour, setFont), the music (setMusic, setMusicLevel, setMusicStart, ids from list_library), the narrator, where a shot or an element sits (placeShot, placeElement) and the format (setFormat). Sound effects play as the film declares them; no edit changes one. Shots count from one. Returns a turn to apply, which the Studio performs.

find_components

ChatGPT
Search the component catalogue by word, for a shot to add. Each result names its text fields and when to use it.

find_layers

ChatGPT
Find layers by shot number, by a word shown on screen, or by component. With no filter, lists the whole film.

get_brand

ChatGPT
Read the project brand: its three colours (ground, accent, ink), its typeface and whether it has a logo. Also lists the typefaces the export draws.

get_project

ChatGPT
Read one project: its brief, its Studio revision, shot count, length, frame shape, template and transitions. Call it before any edit to learn baseRevision.

get_render

ChatGPT
Read the status of one export, and why it failed when it did.

get_render_download

ChatGPT
Return a download link for a finished export. The link expires after five minutes; ask again rather than storing it.

get_storyboard_job

ChatGPT
Read where the newest storyboard generation of a project has got to.

get_turn

ChatGPT
Read whether a proposed turn was applied, and at which revision. A shot-change or composed-screen turn also returns its typed operations, a template turn its film and a film turn its edits. stale: true means it will never be applied (the film moved on, or a day passed): read get_project and propose it again.

inspect_layer

ChatGPT
Read one layer and its controls, with their types, bounds, variants and current values. On a template film each control carries its place: one marked editable: false is refused by propose_edits whatever the value, and its fixedBecause says where it changes instead (the brand name through edit_film setBrandName).

inspect_template

ChatGPT
Read one template in detail before it is used: each shot, what it shows and how long it lasts, the lines and pictures it takes with their limits, its looks, and the products it does not suit.

list_assets

ChatGPT
List the files this project owns by persistent id: pictures, clips, tracks and fonts. Only a picture can be placed with setAsset.

list_backgrounds

ChatGPT
List the backgrounds a shot can take, by the id setGround names.

list_library

ChatGPT
List the music a film may take (id, name, mood: calm, upbeat, cinematic, corporate, playful, and seconds), to name one in edit_film, or the sound effects of one family: informational only, since edit_film no longer places or swaps one (Pierre, 2026-10-22).

list_projects

ChatGPT
List the projects this account can open, most recently changed first, with whether each is its own or shared with it. Returns at most 100; omitted counts the rest.

list_renders

ChatGPT
List the exports of one project, newest first.

list_templates

ChatGPT
List the Library templates a film can be made from: what each tells and for which kind of product, its length, formats, preview and the places the product fills. Look at them before choosing one.

list_transitions

ChatGPT
List the transitions the export draws between two shots, and the three lengths, for setTransition.

preview_component

ChatGPT
Return the shipped preview image of one catalogue component. Demonstration content, not a frame of this film.

propose_edits

ChatGPT
Validate one batch of typed edits (text, numbers, colours, variants, pictures by asset id, backgrounds, shot lengths, transitions) against the film at baseRevision. Saves nothing: returns a turn to apply.

propose_structure

ChatGPT
Validate adding a shot at the end or removing one; shots are never reordered, the order is the template's. Saves nothing: returns a turn. The Studio performs it once applied: at once when it is open, otherwise when the person next opens the project.

rebuild_screen_from_capture

ChatGPT
Experimental, off unless switched on. Rebuild the first screen of the project's own site from its capture into a version 2 composed screen: up to 96 elements where the page drew them, in live text, pictures as crops, with a default direction (3D rise, focus, cursor clicks, typing, counts, a captured second state). No AI call. assetId: a list_assets picture named site-capture. Pass screen, images and seconds to compose_screen; edit beats or drop elements, never add words.

retry_render

ChatGPT
Render a failed export again from the film as it stands, on the original export settings. Costs credits only on success. A second call returns the retry already started.

set_brand_style

ChatGPT
Set the project brand colours and typeface used by generation and adaptation. Only the fields given change: the others keep the stored brand, or the product default when none is stored. Keeps the logo. Does not repaint shots already on the timeline.

set_brief

ChatGPT
Set what the film is about: product, audience, goal, call to action, workflow and target length. Write product as "Name: what it does" ("Ledgerly: bookkeeping for freelancers"): the name before the colon signs a template film, and the call to action fills its last shot. Used by the next generation and by use_template.

start_render

ChatGPT
Render the saved Studio film to MP4. Costs credits, charged only if the export succeeds. Refused if the film changed since baseRevision. A second call for the same film returns the export already started.

start_storyboard_job

ChatGPT
Write a film from a prompt, using the project's screenshots and brief. Spends the AI generation budget, and a new film's credit cost when the product charges for films (charged only once it is written). A second call while one runs joins it. The film is offered in the Studio when the project is opened.

use_template

ChatGPT
Validate a film made from a Library template: a look, every required line of inspect_template, pinned project pictures. No AI generation: you write every line, within inspect_template's limits; fromBrief places (name, address, call to action) come from set_brief. Replaces the whole film: ask the person first if the film already has shots. Returns a turn to apply, which the Studio performs. baseRevision is 0 for a project never opened in the Studio.

Capabilities

InteractiveWrites

Example Prompts

Click any prompt to copy it.

Similar Apps

Listed on MCP App Store

Listed on MCP App Store

Add this badge to your site or README. It links to this page.

HTML

<a href="https://mcp-app-store.keenethics-labs.com/app/scenivia"><img src="https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg" alt="Listed on MCP App Store" width="220" height="44"></a>

Markdown

[![Listed on MCP App Store](https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg)](https://mcp-app-store.keenethics-labs.com/app/scenivia)