Click on "Install 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., "@Agnes MCP Server Generate an image of a futuristic city at sunset, 1920x1080"
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.
An MCP (Model Context Protocol) server that wraps the Agnes AI image and video generation APIs, enabling AI assistants (Claude, Cursor, VS Code, etc.) to generate images and videos via standardized tool calls.
Image Generation — Text-to-image and image-to-image via generate_image
Video Generation — Text-to-video and image-to-video via generate_video + get_video
Two Transport Modes — stdio (local) and Streamable HTTP (remote)
Auto Download — Optionally save generated media to disk automatically
Blocking & Async — wait=true for synchronous generation, or poll with get_video
Related MCP server: ImaginePro MCP Server
cd agnes-mcp-server
npm install
npm run buildexport AGNES_API_KEY="sk-agnes-your-api-key-here"Stdio mode (default, for local MCP clients like Claude Desktop, Cursor):
npm run start
# or
node agnes-mcp-server.cjsStreamable HTTP mode (for remote MCP clients):
MCP_TRANSPORT=http npm run start:http
# Default port: 3100 (configurable via HTTP_PORT)Add this to your MCP client configuration (e.g., .mcp.json):
{
"mcpServers": {
"agnes": {
"command": "node",
"args": ["agnes-mcp-server.cjs"],
"env": {
"AGNES_API_KEY": "sk-agnes-your-api-key-here"
}
}
}
}Tool | Description |
| Generate or edit an image. Supports text-to-image and image-to-image modes. |
| Submit a video generation task. Set |
| Query video generation status and optionally download the result. |
generate_image
Parameter | Type | Required | Default | Description |
| string | Yes | — | Text description of the desired image |
| string | No |
| Image dimensions (e.g., |
| string[] | No | — | Input images (local file paths or base64 data URIs) |
|
| No |
| Output format when no |
| string | No | — | Directory to download the image file |
generate_video
Parameter | Type | Required | Default | Description |
| string | Yes | — | Text description of the desired video |
| string | string[] | No | — | Input image(s) for image-to-video mode |
| string | No | — | Generation mode (e.g., |
| number | No | — | Video width |
| number | No | — | Video height |
| number | No | — | Number of frames |
| number | No | — | Frames per second |
| number | No | — | Random seed for reproducibility |
| string | No | — | Things to exclude from the video |
| string | No | — | Directory to download the video file |
| boolean | No |
| Block until generation completes |
get_video
Parameter | Type | Required | Default | Description |
| string | Yes | — | Video ID returned by |
| string | No | — | Task ID (fallback for querying) |
| string | No | — | Directory to download the video file |
Variable | Required | Default | Description |
| Yes | — | Your Agnes API key (starts with |
| No |
| Transport mode: |
| No |
| HTTP server port (only in |
| No | — | Root directory for auto-downloaded media |
agnes-mcp-server/
├── src/
│ ├── app/ # Server entry points (stdio, http, bundle)
│ ├── client/ # HTTP client wrapper
│ ├── core/ # Core MCP setup and registry
│ ├── module/
│ │ ├── image/ # Image generation service
│ │ └── video/ # Video generation service
│ ├── providers/ # API providers (AgnesClient, image, video)
│ ├── tools/ # MCP tool implementations
│ └── types/ # Shared types and result formatters
├── agnes-mcp-server.cjs # Bundled entry point
├── agnes-bundle.cjs # Standalone bundle
├── SKILL.md # MCP skill definition
├── TOKEN_AUTH.md # Authentication documentation
└── package.json# Watch mode (TypeScript → Node)
npm run dev
# Build TypeScript
npm run build
# Create standalone bundle
npm run bundle
# Run smoke test
npm run testA standalone test script is included for verifying the image-to-video API flow:
npx tsx test-video-image.ts <local-image-path> <api-key>
# or
IMAGE_PATH=test.jpg AGNES_API_KEY=sk-xxx npx tsx test-video-image.tsMIT
Generate or edit an image using the Agnes AI image generation API. Supports text-to-image and image-to-image modes.
| Name | Required | Description | Default |
|---|---|---|---|
| size | No | 1024x1024 | |
| format | No | url | |
| images | No | ||
| prompt | Yes | ||
| outputDir | No |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations, the description must disclose behavioral traits. It mentions two modes but does not explain prerequisites (e.g., requiring images parameter for editing), return format, or other side effects. This is insufficient for a tool with 5 parameters.
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 concise with two sentences and no redundancy. However, it sacrifices necessary detail in favor of brevity. It is well-structured but lacks depth.
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?
Given 5 parameters, no output schema, and no annotations, the description is incomplete. It does not address return values, parameter roles, or typical use cases, making it inadequate for complex interactions.
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 has 0% description coverage, and the description adds almost no meaning beyond stating modes. Parameters like size, format, images, and outputDir are not explained, leaving the agent with no guidance on how to use them.
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 purpose: generate or edit an image using a specific API. It mentions two modes (text-to-image and image-to-image), which distinguishes it from sibling tools that handle 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?
The description implies usage for image generation/editing but does not explicitly guide when to use this tool versus siblings like generate_video or get_video. No exclusions or alternatives are mentioned.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Submit a video generation task. By default returns a videoId immediately for progress polling with get_video. Set wait=true to block until generation completes. Optionally provide outputDir to auto-download the video when completed.
| Name | Required | Description | Default |
|---|---|---|---|
| mode | No | ||
| seed | No | ||
| wait | No | ||
| image | No | ||
| width | No | ||
| height | No | ||
| prompt | Yes | ||
| outputDir | No | ||
| frame_rate | No | ||
| num_frames | No | ||
| negative_prompt | No |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations, the description carries full burden. It discloses the default async behavior (returns videoId), the blocking option, and auto-download. However, it omits potential side effects, authentication, rate limits, or failure handling.
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 concise sentences, each adding value. Front-loaded with the primary action, then options. No redundant information.
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?
Given 11 parameters, no output schema, and no annotations, the description is incomplete. It covers key workflow but leaves many parameters unexplained, forcing the agent to guess or rely on external knowledge.
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 0%, so description must compensate. It only explains wait and outputDir, leaving 9 parameters (prompt, mode, seed, image, width, height, frame_rate, num_frames, negative_prompt) undocumented. This is insufficient for an agent to use the tool effectively.
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 identifies the tool's purpose: submitting a video generation task. It distinguishes from siblings by mentioning videoId polling (related to get_video) and implies difference from generate_image.
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 explicit guidance on using wait=false (default) for polling vs wait=true for blocking, and mentions outputDir for auto-download. It does not explicitly exclude scenarios but offers clear context.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Get the current status of a video generation task and optionally download it. Accepts videoId (required) and taskId (optional). Returns progress percentage, status (queued/in_progress/completed/failed), and download URL when completed. Provide outputDir to auto-download the video file when generation is complete.
| Name | Required | Description | Default |
|---|---|---|---|
| taskId | No | ||
| videoId | Yes | ||
| outputDir | No |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations, the description fully covers behavior: it's a read operation returning status and download URL, with optional auto-download via outputDir. It does not mention side effects, but 'Get' implies non-destructive. The description adds context beyond minimal.
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 concise sentences: first frames purpose, second lists parameters and returns, third notes optional auto-download. No redundant information; every sentence 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?
Given no output schema and no annotations, the description covers inputs, outputs (progress, status, download URL), and optional behavior. It lacks mention of error cases, prerequisites (e.g., needing to have called generate_video first), or authentication needs, but is fairly complete for a simple status-checking tool.
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 schema has 0% coverage (no parameter descriptions). The description explains videoId (required), taskId (optional for identifying task), and outputDir (for auto-download). This adds meaning beyond the raw schema, though formats and constraints are not detailed.
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 retrieves the status of a video generation task and optionally downloads it. The verb 'Get' and resource 'video generation task' are specific. Siblings 'generate_image' and 'generate_video' are distinct, so no confusion.
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 use after initiating a generation task but does not explicitly state when to use this tool versus alternatives. No mention of prerequisites or when not to use it. The context of siblings suggests usage, but explicit guidance is missing.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Each tool targets a distinct operation: image generation, video generation submission, and video status polling/download. No overlapping functionality, making selection unambiguous.
All tool names follow a consistent verb_noun pattern in snake_case (generate_image, generate_video, get_video), with no deviations or mixed conventions.
Three tools cover the essential workflows for image and video generation without unnecessary bloat, fitting well within the ideal 3-15 range.
Core generation and status retrieval are covered. Missing delete or list operations for generated content, but these are minor gaps given the server's likely purpose.
Unclaimed servers have limited discoverability.
If you are the server author, to access and configure the admin panel.
Generate AI images, videos, music, SFX & speech in any AI assistant. Results appear inline in chat.
Generate images, videos, voiceovers, and captions from a chat prompt.
Generate images, video, and audio with Glif's media-generation agent
AI image, video & audio generation tools for AI agents, powered by Varoriya.
We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/zssty2010/agnes-mcp-server'
If you have feedback or need assistance with the MCP directory API, please join our Discord server
