Provides tools for creating, managing, and remixing AI-generated videos using OpenAI's Sora API, including video generation, status monitoring, downloading completed videos, and applying targeted adjustments to existing videos.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@Sora2 MCP create a 5-second video of a cat riding a skateboard in a cyberpunk city"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
昭和レトロCM風デモ動画を生成しました！
Video ID: video_68e4f09ad79c8198a43196de75bbe1b10170aa9f31b5a6a2
🎨 スタイル: 1960年代 日本のTV CM風
⏱️ 長さ: 4秒
📐 解像度: 1280x720
🤖 モデル: sora-2
このデモは、Sora2 MCPを使って自動生成された昭和初期カラーTV時代のレトロなコマーシャル風の動画です。
The Universal MCP Server exposes tools for OpenAI's Sora API, enabling programmatic creation, management, and remixing of AI-generated videos. Designed for prompt-first usage in MCP-compatible clients.
Related MCP server: Sora 2 MCP Server
Node.js 18+
Set SORA2_MCP_API_KEY (or OPENAI_API_KEY) in your environment
Obtain your OpenAI API key from OpenAI Platform
Ensure your account has access to the Sora API (currently in limited beta)
For more information, see the Sora API documentation
cd /path/to/sora2-mcp
npm i
npm run buildUse this one-liner (replace with your real values):
claude mcp add sora2-mcp -s user -e SORA2_MCP_API_KEY="sk-your-real-key" -- npx sora2-mcpTo remove:
claude mcp remove sora2-mcpNote: This repository does not include .cursor/mcp.json. Configure Cursor via its UI settings, or create .cursor/mcp.json in your client workspace (do not commit it here):
{
"mcpServers": {
"sora2-mcp": {
"command": "npx",
"args": ["sora2-mcp"],
"env": {
"SORA2_MCP_API_KEY": "sk-your-real-key"
},
"autoStart": true
}
}
}Install via CLI:
code --add-mcp '{"name":"sora2-mcp","command":"npx","args":["sora2-mcp"],"env":{"SORA2_MCP_API_KEY":"sk-your-real-key"}}'Or configure in your VS Code settings under MCP servers.
Similar to VS Code, use the Insiders binary:
code-insiders --add-mcp '{"name":"sora2-mcp","command":"npx","args":["sora2-mcp"],"env":{"SORA2_MCP_API_KEY":"sk-your-real-key"}}'Add to your Claude Desktop MCP configuration file:
macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%\Claude\claude_desktop_config.json
{
"mcpServers": {
"sora2-mcp": {
"command": "npx",
"args": ["sora2-mcp"],
"env": {
"SORA2_MCP_API_KEY": "sk-your-real-key"
}
}
}
}Restart Claude Desktop after saving.
In LM Studio's MCP settings:
Command: npx
Args: ["sora2-mcp"]
Environment Variables:
SORA2_MCP_API_KEY = sk-your-real-key
Configure in Goose's MCP settings:
Type: STDIO
Command: npx
Args: sora2-mcp
Enabled: true
Add environment variable SORA2_MCP_API_KEY in your shell or Goose config.
Example ~/.config/opencode/opencode.json:
{
"$schema": "https://opencode.ai/config.json",
"mcp": {
"sora2-mcp": {
"type": "local",
"command": ["npx", "sora2-mcp"],
"env": {
"SORA2_MCP_API_KEY": "sk-your-real-key"
},
"enabled": true
}
}
}Add a new MCP server in Qodo Gen settings and paste the standard JSON config:
{
"name": "sora2-mcp",
"command": "npx",
"args": ["sora2-mcp"],
"env": {
"SORA2_MCP_API_KEY": "sk-your-real-key"
}
}Configure in Windsurf's MCP settings. Reuse the standard config above with command npx and args ["sora2-mcp"].
Add the following to your Codex TOML configuration.
Example (Serena reference):
[mcp_servers.serena]
command = "uvx"
args = ["--from", "git+https://github.com/oraios/serena", "serena", "start-mcp-server", "--context", "codex"]This server (minimal):
[mcp_servers.sora2-mcp]
command = "npx"
args = ["sora2-mcp"]
# Optional environment variables:
# SORA2_MCP_API_KEY = "sk-your-real-key"
# MCP_NAME = "sora2-mcp"SORA2_MCP_API_KEY (or OPENAI_API_KEY): Your OpenAI API key with Sora access
MCP_NAME: Server name override (default: sora2-mcp)
create_videoStart a new video generation job with Sora. Returns a job object with status. Poll get_video_status or use webhooks to monitor completion.
Inputs:
prompt (string, required): Text description of the video. Be specific about shot type, subject, action, setting, and lighting for best results.
model (string, optional): Model to use. "sora-2" (faster for iteration) or "sora-2-pro" (higher quality for production). Default: "sora-2"
size (string, optional): Video resolution. Options: "1280x720", "1920x1080", "720x1280", "1080x1920". Default: "1280x720"
seconds (string, optional): Video duration. Options: "4", "8", "12". Default: "8"
input_reference (string, optional): Base64-encoded image (JPEG, PNG, or WebP) to use as the first frame. Must match target resolution.
Outputs: JSON object with id, status (queued, in_progress, completed, failed), progress, and other metadata.
get_video_statusRetrieve the current status and progress of a video generation job.
Inputs:
video_id (string, required): The video job ID returned from create_video
Outputs: JSON object with status, progress percentage, and metadata.
download_videoDownload the completed video file (MP4), thumbnail (WebP), or spritesheet (JPEG). Only works when status is completed. Returns base64-encoded binary data.
Inputs:
video_id (string, required): The video job ID
variant (string, optional): What to download. Options: "video" (MP4), "thumbnail" (WebP), "spritesheet" (JPEG). Default: "video"
Outputs: JSON object with video_id, variant, size_bytes, data_base64, and usage notes.
list_videosList all video jobs with pagination support. Returns metadata for enumeration, dashboards, or housekeeping.
Inputs:
limit (number, optional): Number of videos to return. Default: 10, Max: 100
after (string, optional): Cursor for pagination to fetch the next page
order (string, optional): Sort order by creation date. Options: "asc", "desc". Default: "desc"
Outputs: JSON object with data array of video objects and pagination metadata.
delete_videoDelete a video from OpenAI's storage. This action is permanent.
Inputs:
video_id (string, required): The video job ID to delete
Outputs: Confirmation JSON object.
remix_videoCreate a new video by remixing an existing completed video with targeted adjustments. Preserves structure and composition while applying modifications.
Inputs:
video_id (string, required): The ID of the completed video to use as the base
prompt (string, required): Description of the change to apply. Keep it focused on a single, well-defined adjustment for best results.
Outputs: JSON object with new job id and status.
For detailed input/output schemas, see src/index.ts.
{
"name": "create_video",
"arguments": {
"prompt": "Wide tracking shot of a teal coupe driving through a desert highway, heat ripples visible, hard sun overhead.",
"model": "sora-2-pro",
"size": "1280x720",
"seconds": "8"
}
}Response:
{
"id": "video_68d7512d07848190b3e45da0ecbebcde004da08e1e0678d5",
"object": "video",
"created_at": 1758941485,
"status": "queued",
"model": "sora-2-pro",
"progress": 0,
"seconds": "8",
"size": "1280x720"
}Then poll with:
{
"name": "get_video_status",
"arguments": {
"video_id": "video_68d7512d07848190b3e45da0ecbebcde004da08e1e0678d5"
}
}Once status is "completed", download with:
{
"name": "download_video",
"arguments": {
"video_id": "video_68d7512d07848190b3e45da0ecbebcde004da08e1e0678d5",
"variant": "video"
}
}401 authentication errors: Check that SORA2_MCP_API_KEY (or OPENAI_API_KEY) is set correctly and that your account has Sora API access.
Ensure Node 18+: Run node -v to verify your Node.js version.
Local runs: After building with npm run build, test locally with npx . or node build/index.js in the project directory.
Inspect publish artifacts: Run npm pack --dry-run to see what files will be published.
Video generation takes time: Depending on model and resolution, rendering may take several minutes. Use polling or webhooks to monitor progress efficiently.
Download URLs expire: Video download URLs are valid for a maximum of 24 hours. Save files promptly to your own storage.
Content restrictions: The API enforces guardrails (no copyrighted characters, no real people, etc.). Ensure prompts and inputs comply.
Always use CANONICAL_ID (sora2-mcp) for identifiers and keys.
Use CANONICAL_DISPLAY (Sora2 MCP) only for UI labels.
Do not mix different names across clients.
Context | Value |
npm package name |
|
Binary name |
|
MCP server name (SDK metadata) |
|
Env default |
|
Client registry key |
|
UI label |
|
Remove any old entries (e.g., legacy display names like "Sora2") from your MCP configuration and re-add with "sora2-mcp" as the key.
Ensure global .mcp.json or client registries only use "sora2-mcp" for keys.
Cursor: Configure in the UI only. This project does not include .cursor/mcp.json.
Correct:
{
"mcpServers": {
"sora2-mcp": {
"command": "npx",
"args": ["sora2-mcp"]
}
}
}Incorrect:
{
"mcpServers": {
"Sora2": {
"command": "npx",
"args": ["sora2-mcp"]
}
}
}Using inconsistent keys like "Sora2" will conflict with "sora2-mcp".
MIT
Start a new video generation job with Sora. Returns a job object with status. Poll get_video_status or use webhooks to monitor completion.
| Name | Required | Description | Default |
|---|---|---|---|
| size | No | Video resolution (width x height) | 1280x720 |
| model | No | Model to use. 'sora-2' is faster for iteration; 'sora-2-pro' is higher quality for production. | sora-2 |
| prompt | Yes | Text description of the video to generate. Be specific about shot type, subject, action, setting, and lighting for best results. | |
| seconds | No | Video duration in seconds | 8 |
| input_reference | No | Optional: Base64-encoded image to use as the first frame (JPEG, PNG, or WebP). Must match target video resolution. |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations, the description carries the full burden. It discloses the async nature (returns a job object, monitor via status/webhooks) and hints at the workflow. This goes beyond basic, though it omits potential details like rate limits, which are not essential for a simple create operation.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Three sentences, all informative and non-redundant. Front-loaded with the primary purpose, followed by return value and monitoring guidance. No wasted words.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a create tool with 5 params and no output schema, the description sufficiently covers the purpose, async workflow, and how to monitor completion. It could elaborate on response fields, but get_video_status is mentioned for that. Overall, it is well-rounded for its complexity.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema description coverage is 100% with detailed descriptions for all 5 parameters. The description adds no parameter-specific guidance beyond the schema, so the baseline of 3 is appropriate.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states 'Start a new video generation job with Sora', which is a specific verb+resource. It distinguishes from siblings (delete_video, get_video_status, download_video, list_videos, remix_video) by focusing on creation, and adds the return of a job object.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
Provides clear context as the entry point for video creation and explicitly directs the user to poll get_video_status or use webhooks for monitoring. However, it does not explicitly state when not to use this tool versus alternatives, so it lacks explicit exclusions.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Delete a video from OpenAI's storage. This action is permanent.
| Name | Required | Description | Default |
|---|---|---|---|
| video_id | Yes | The video job ID to delete |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
The description discloses the key behavioral trait that deletion is permanent, which is crucial for a destructive operation, but lacks additional context such as error behavior or permissions. Since no annotations are provided, the description carries the burden, but 'permanent' provides some transparency.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is two sentences with no fluff, front-loading the action and adding a critical warning about permanence. Every word earns its place.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a simple delete tool with one parameter and no output schema, the description is adequately complete, covering the action, scope, and permanence. It could add details about error handling or idempotency, but these are not necessary for such a straightforward operation.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
The input schema already fully documents video_id as 'The video job ID to delete', and the description does not add additional parameter semantics. With 100% schema coverage, the description adds no extra meaning beyond what the schema provides.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the tool deletes a video from OpenAI's storage with the verb 'Delete' and resource 'video', and it distinguishes from sibling tools (create, get, download, list, remix) by being the only delete operation.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description provides no explicit guidance on when to use this tool versus alternatives or any prerequisites. It only states the action without context for when deletion is appropriate or any exclusions.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Download the completed video file (MP4), thumbnail (WebP), or spritesheet (JPEG). Only works when status is 'completed'. Returns base64-encoded binary data.
| Name | Required | Description | Default |
|---|---|---|---|
| variant | No | What to download: 'video' (MP4), 'thumbnail' (WebP), or 'spritesheet' (JPEG) | video |
| video_id | Yes | The video job ID |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations provided, the description carries the full burden. It does disclose the return type (base64-encoded binary data) and the precondition, which is useful. However, it does not specify behavior on failure, such as what error occurs if status is not completed, or whether the download is read-only. These are minor gaps for a download operation.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is two concise sentences. The first sentence states the core action and variants; the second provides the critical condition and return format. There is no filler or redundant content.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a simple download tool with only two parameters, this description is complete. It covers the resource types, the precondition, and the return format. The absence of an output schema is compensated by the explicit mention of base64-encoded data. Sibling tools are clearly different in purpose, so no additional context is needed.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema coverage is 100%, so the baseline is 3. The description goes beyond by mapping each variant to its format (video=MP4, thumbnail=WebP, spritesheet=JPEG) and adds the completed-status condition for video_id. This enriches the schema's terse definitions.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description uses the specific verb 'Download' and clearly identifies the resources (video, thumbnail, spritesheet) with their formats (MP4, WebP, JPEG). This completely distinguishes it from sibling tools like delete_video or create_video.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
It states the explicit precondition 'Only works when status is completed', which guides the agent on when it is appropriate to call. It does not explicitly name alternatives like get_video_status for checking status, but the condition implies the need to check first.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Retrieve the current status and progress of a video generation job. Status values: queued, in_progress, completed, failed.
| Name | Required | Description | Default |
|---|---|---|---|
| video_id | Yes | The video job ID returned from create_video |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations provided, the description carries full burden. It discloses possible status values and mentions progress, which is useful. However, it does not explicitly state that this is a read-only operation, how errors are handled, or what the response structure will look like. Moderate transparency for a simple status-check tool.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is two concise sentences, front-loaded with the action and resource. The status value enumeration is informative and adds value without redundancy.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a simple single-parameter status tool, the description covers the core operation and enumerates status values. However, without an output schema or annotations, it does not clarify what 'progress' means or describe the response format, leaving some ambiguity. It is still largely complete for its simplicity.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema coverage is 100%, and the schema description already states that video_id is returned from create_video. The tool description adds no additional parameter semantics beyond what the schema provides, so the baseline of 3 applies.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description uses a specific verb ('retrieve') and clearly identifies the resource ('current status and progress of a video generation job'). It distinguishes itself from sibling tools by focusing on status polling, and the enumeration of status values reinforces the purpose.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description implies usage after creating a video job but does not explicitly state when to use this tool or exclude alternatives. No mention of 'use after create_video' or comparison with sibling tools, so guidance is only implicit.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
List all video jobs with pagination support. Returns metadata for enumeration, dashboards, or housekeeping.
| Name | Required | Description | Default |
|---|---|---|---|
| after | No | Cursor for pagination. Use this to fetch the next page. | |
| limit | No | Number of videos to return (default 10, max 100) | |
| order | No | Sort order by creation date | desc |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations, the description discloses pagination support, the 'all' scope, and that it returns metadata. It does not describe the metadata fields, default sort order, or any edge cases (e.g., empty pages), leaving some behavior implicit.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Two sentences: the first states the action and scope, the second states the purpose. Every word earns its place; no filler or repetition of schema details.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a simple list tool with no output schema, the description adequately covers purpose and pagination. It could mention the default ordering or the exact metadata fields returned, but the schema covers parameter details, making this sufficient for basic use.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
The input schema provides thorough descriptions for all three parameters (after, limit, order) with default values and max constraints, so the description does not need to repeat them. The description's mention of pagination aligns with the 'after' cursor parameter, adding minimal extra context.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description uses a specific verb ('List') and resource ('video jobs'), and 'with pagination support' further clarifies scope. This clearly distinguishes it from siblings like create_video, delete_video, and get_video_status, which perform different operations.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
It gives explicit usage context: 'for enumeration, dashboards, or housekeeping.' However, it does not explicitly contrast with alternatives like get_video_status, which targets a single video's status, so the guidance is clear but lacks exclusions.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Create a new video by remixing an existing completed video with targeted adjustments. Preserves structure and composition while applying modifications.
| Name | Required | Description | Default |
|---|---|---|---|
| prompt | Yes | Description of the change to apply. Keep it focused on a single, well-defined adjustment for best results. | |
| video_id | Yes | The ID of the completed video to use as the base for remixing |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations, the description carries the burden of behavioral disclosure. It adds a meaningful behavioral trait ('Preserves structure and composition'), but does not mention permissions, async behavior, return values, or side effects on the original video. Still, it offers more than a tautological statement.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is two sentences, front-loaded with the primary action and scope. The second sentence adds a valuable behavioral note without redundancy. No filler or repetition of schema fields.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
The tool is a mutation with no output schema and no annotations. The description explains the main behavior but omits details about what the caller receives (e.g., video ID, status), whether processing is synchronous, and any specific requirements beyond a completed video. It is minimally sufficient but leaves key operational gaps.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema description coverage is 100%, so the baseline is 3. The description adds some context for the 'prompt' parameter ('targeted adjustments') but adds no new meaning for 'video_id' beyond the schema. It does not compensate further.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the tool's function: creating a new video by remixing an existing completed video with targeted adjustments. It explicitly mentions the base resource ('existing completed video') and the action ('remixing'), distinguishing it from siblings like create_video, which would start from scratch.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description provides clear context: use when you have an existing completed video and want to apply targeted modifications while preserving structure. It implicitly distinguishes from create_video by requiring a completed video, though it does not explicitly name alternatives or exclusions.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
v1.0.0create_videodelete_videodownload_videoget_video_statuslist_videosremix_videoScored across 6 tools
Each tool has a clearly distinct purpose: create, status check, download, list, delete, and remix. There is no overlap between them, and even the two retrieval tools (get_video_status and download_video) are cleanly separated by their output.
All tool names follow a consistent verb_noun pattern in snake_case (e.g., create_video, get_video_status, download_video). The only minor variance is list_videos being plural, which is natural for a collection operation and does not break the pattern.
Six tools is well-scoped for a video generation service. The set covers the full job lifecycle (create, poll, download, list, delete) plus an additional remix feature, without unnecessary redundancy or bloat.
The core lifecycle is covered, but there is no way to cancel an in-progress job or fetch full metadata for a single video without listing all jobs. These are minor gaps that agents can work around, so the surface is mostly complete.
Create and manage cinematic AI video renders through the Future Video Studio Agent API.
Generate, edit, and render videos via the Shotstack video editing API.
LLM chat, text tools, image generation, editing, batch image jobs, and asynchronous video generation
One API for 100+ AI video, image, music and speech models.
