MCP App Store
by KeenEthics
Media.io icon

Media.io MCP Server for ChatGPT

By Wondershare

Overview

Create images and videos with model selection across a static catalog of 119 model-operation entries. Generate from text, images or declared mixed references, transfer motion from a video, or create a music video from an image and audio. Reuse private Media.io drive files, view paginated history cards, monitor tasks and display finished media. Check existing credits and estimate costs when requested. Generation consumes existing account credits and entitlements; models and options are limited to the current tool schemas.

How to connect

  1. ChatGPT: Open the Media.io listing in the ChatGPT app directory and choose Connect. ChatGPT listing Media.io exposes 19 tools here.

Tools

complete_upload

ChatGPT
Step 3 of uploading a local file. Call it only after the PUT to upload_url returned a 2xx, and never after a rapid_upload shortcut. The upstream verifies the stored object against the size and hash declared in create_upload, then registers the file in the drive and returns file_id and asset_id. file_id is the value generation parameters take. asset_id is a drive bookkeeping id and is rejected as a generation input. It is idempotent: calling it again for an already completed upload returns the same result without creating a duplicate. UPLOAD_NOT_READY means the bytes have not landed in storage yet — finish or retry the PUT and call this again. UPLOAD_EXPIRED means the ticket is gone; start over with create_upload.

create_upload

ChatGPT
Step 1 of uploading a local file into the user's Media.io drive. It never receives file bytes: it takes the file metadata plus two SHA-1 digests and returns a presigned URL that the caller PUTs the bytes to. Compute content_hash as the SHA-1 of the whole file and pre_hash as the SHA-1 of its first 1 MiB (for a file of 1 MiB or less the two are identical). When the response has rapid_upload: true and state: "completed", the drive already had that exact content. The upload is done — do not PUT anything and do not call complete_upload. Otherwise PUT the raw file bytes to upload_url using upload_method, sending every entry of upload_headers exactly as given. Do not add, drop, rename or reorder those headers, and do not rewrite the URL, or the storage will reject the signature. Then call complete_upload with the returned upload_id. If the PUT does not finish before expires_at, call create_upload again for a fresh URL. file_id is the value generation parameters take. asset_id is a drive bookkeeping id and is rejected as a generation input.

estimate_credits

ChatGPT
Use this when the user asks about cost or requests a credit estimate for a declared generation action and model. Supply request.operation and its complete request.generation using the same inputs and options as generation. This is read-only, never submits a task, and is not a prerequisite for generation. Estimates are not final charge receipts.

get_account

ChatGPT
Reads existing account entitlements and available credit balance. It does not render a card. Use get_account_card only when the user asks to see their account; this query is not required before normal generation.

get_account_card

ChatGPT
Use this when the user asks to see their connected account, credit balance or membership. Returns the account nickname, email, existing balance and membership status and displays them in a card. The card has no buttons or navigation links. This does not offer purchases or change entitlements. Do not repeat the rendered data in prose.

get_generation

ChatGPT
Use this to inspect or wait for existing generation tasks by task_id. Returns status and available outputs as data without rendering a card or creating tasks. The call holds until every task is terminal or wait_ms elapses, so it is normal for one call to take many seconds. When it returns with all_terminal false, wait poll_after_ms and call it again with the tasks that are still running. Decide whether a task has finished from status_label and terminal; do not parse the numeric status. For a user-requested generation batch, wait until every task is terminal, then call get_generation_card once to display the results. If the host cannot render cards, deliver available outputs[].url only through media or link capabilities supported by that host, preserving each signed URL verbatim. Do not re-encode signed URLs or download media to bypass display restrictions; save a local copy only when the user requests it and the host permits it.

get_generation_card

ChatGPT
Shows the user the finished results of generation tasks: images, video and audio are rendered inline with a save link, and a failed task shows its reason. Same data as get_generation, except this one renders. Call it exactly once per request, after get_generation reports all_terminal true. It is not a progress indicator — calling it on running tasks shows the user a card that says nothing useful. The card already shows the media, so afterwards say at most one short sentence and do not describe what the user can already see.

get_generation_model_guide

ChatGPT
Read exact field paths, required inputs, defaults, allowed values, source-media limits, conditional option combinations and validated example templates for one model already declared by a generation tool. If its ID is unknown, first use list_generation_models with the intended action. Examples require real prompts, authorized file IDs and a fresh stable request ID. This local guide never executes a request or adds a model. Generation tools remain callable directly with valid arguments.

image_audio_to_video

ChatGPT
Use this when the user wants to create a music video from an image and audio. Select request.model and follow that model's exact request.inputs and request.options. Map requested aspect ratio, pixel size, resolution and duration to the declared request.options, including settings written in the prompt. Defaults apply only to unspecified settings. Resolve unsupported requirements before submission. All callable models and settings are declared in this schema; other models and operations are unavailable. Creates one task and normally consumes existing account credits. Submit directly when requested; no prior estimate or per-task credit approval is required. Preserve source files. Poll get_generation and deliver with get_generation_card. Never resubmit or change the request ID after RESULT_UNKNOWN; inspect list_generations first.

image_to_image

ChatGPT
Use this when the user wants to create an image from images. Select request.model and follow that model's exact request.inputs and request.options. Map requested aspect ratio, pixel size, resolution and duration to the declared request.options, including settings written in the prompt. Defaults apply only to unspecified settings. Resolve unsupported requirements before submission. All callable models and settings are declared in this schema; other models and operations are unavailable. Creates one task and normally consumes existing account credits. Submit directly when requested; no prior estimate or per-task credit approval is required. Preserve source files. Poll get_generation and deliver with get_generation_card. Never resubmit or change the request ID after RESULT_UNKNOWN; inspect list_generations first.

image_to_video

ChatGPT
Use this when the user wants to animate an image. Select request.model and follow that model's exact request.inputs and request.options. Map requested aspect ratio, pixel size, resolution and duration to the declared request.options, including settings written in the prompt. Defaults apply only to unspecified settings. Resolve unsupported requirements before submission. All callable models and settings are declared in this schema; other models and operations are unavailable. Creates one task and normally consumes existing account credits. Submit directly when requested; no prior estimate or per-task credit approval is required. Preserve source files. Poll get_generation and deliver with get_generation_card. Never resubmit or change the request ID after RESULT_UNKNOWN; inspect list_generations first.

list_assets

ChatGPT
Lists the media assets in the current subject drive, with pagination, so they can be picked as inputs for a generation task. Pass an entry's file_id into the generation parameter; asset_id is a drive bookkeeping id and is rejected as a generation input. Given an asset_id from a finished task's outputs, filter by it here to get the matching file_id.

list_generation_models

ChatGPT
List exact model IDs, names, aliases and selection defaults already declared by this MCP release. Use when a model ID is unknown or the user asks which models are supported. Filter by the intended generation action and optionally a model name or alias; do not send prompts or conversation text. Honor explicit model and cost constraints first, then scenario rules, membership conditions and ordinary defaults. Read guidance for free recommendations and conditional member defaults; clarify materially different candidates. Read get_generation_model_guide for parameter details. This fixed catalog never creates tasks, estimates credits or enables new models. Status does not guarantee account entitlement or live provider availability.

list_generations

ChatGPT
Lists the generation tasks submitted by the current subject, with pagination. Returns data without displaying a card; use list_generations_card when the user asks to view generation history. After a RESULT_UNKNOWN, use this data tool to inspect recent task pages and their timestamps to determine whether the task was created before deciding to resubmit. The output fields mirror get_generation.

list_generations_card

ChatGPT
Shows a page of the authenticated user's generation history in a card, including task status, timestamps and available media. Use when the user asks to view past or recent generations. Supports the same task ID and status filters and pagination as list_generations, with at most 100 tasks per page. Show the current snapshot immediately, including running and failed tasks; do not wait for every history task to finish or add a separate result card per task. For another page, call this tool with the same filters and the requested page. Afterwards say at most one short sentence. This read-only tool never creates tasks or spends generation credits.

motion_control

ChatGPT
Use this when the user wants to animate an image using a motion video. Select request.model and follow that model's exact request.inputs and request.options. Map requested aspect ratio, pixel size, resolution and duration to the declared request.options, including settings written in the prompt. Defaults apply only to unspecified settings. Resolve unsupported requirements before submission. All callable models and settings are declared in this schema; other models and operations are unavailable. Creates one task and normally consumes existing account credits. Submit directly when requested; no prior estimate or per-task credit approval is required. Preserve source files. Poll get_generation and deliver with get_generation_card. Never resubmit or change the request ID after RESULT_UNKNOWN; inspect list_generations first.

reference_to_video

ChatGPT
Use this when the user wants to create a video from references. Select request.model and follow that model's exact request.inputs and request.options. Map requested aspect ratio, pixel size, resolution and duration to the declared request.options, including settings written in the prompt. Defaults apply only to unspecified settings. Resolve unsupported requirements before submission. All callable models and settings are declared in this schema; other models and operations are unavailable. Creates one task and normally consumes existing account credits. Submit directly when requested; no prior estimate or per-task credit approval is required. Preserve source files. Poll get_generation and deliver with get_generation_card. Never resubmit or change the request ID after RESULT_UNKNOWN; inspect list_generations first.

text_to_image

ChatGPT
Use this when the user wants to create an image from text. Select request.model and follow that model's exact request.inputs and request.options. Map requested aspect ratio, pixel size, resolution and duration to the declared request.options, including settings written in the prompt. Defaults apply only to unspecified settings. Resolve unsupported requirements before submission. All callable models and settings are declared in this schema; other models and operations are unavailable. Creates one task and normally consumes existing account credits. Submit directly when requested; no prior estimate or per-task credit approval is required. Preserve source files. Poll get_generation and deliver with get_generation_card. Never resubmit or change the request ID after RESULT_UNKNOWN; inspect list_generations first.

text_to_video

ChatGPT
Use this when the user wants to create a video from text. Select request.model and follow that model's exact request.inputs and request.options. Map requested aspect ratio, pixel size, resolution and duration to the declared request.options, including settings written in the prompt. Defaults apply only to unspecified settings. Resolve unsupported requirements before submission. All callable models and settings are declared in this schema; other models and operations are unavailable. Creates one task and normally consumes existing account credits. Submit directly when requested; no prior estimate or per-task credit approval is required. Preserve source files. Poll get_generation and deliver with get_generation_card. Never resubmit or change the request ID after RESULT_UNKNOWN; inspect list_generations first.

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/wondershare-media-io"><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/wondershare-media-io)