generate_image
An image from a text prompt, shown inline in the chat. Also saved to disk when AIBUDDY_OUTPUT_DIR is set.
Try: Make a 16:9 hero image of a mountain cabin in winter, warm window light.
AIBuddy MCP for any AI agent
Images, video and music from any AI agent
Connect AIBuddy to Claude, Cursor, Codex or VS Code and generate images, video and music from your prompts, billed to your AIBuddy credits.
Install in under a minute
Pick how you work. Both paths run the same package with npx, so you need Node.js installed and nothing else.
MCP server
Claude Desktop
- Get your AIBuddy API key. It starts with ab_live_. Create a key
-
Open Settings > Developer > Edit Config and paste this into claude_desktop_config.json, replacing ab_live_... with your key:
{ "mcpServers": { "aibuddy": { "command": "npx", "args": ["-y", "github:ThomasWDev/aibuddy-mcp"], "env": { "AIBUDDY_API_KEY": "ab_live_..." } } } } - Restart the client, then try:
Use aibuddy to make a 16:9 image of a lighthouse at dusk in watercolor.
Claude Code
- Get your AIBuddy API key. It starts with ab_live_. Create a key
-
Run this in your terminal, replacing ab_live_... with your key:
claude mcp add aibuddy -e AIBUDDY_API_KEY=ab_live_... -- npx -y github:ThomasWDev/aibuddy-mcp - Restart the client, then try:
Use aibuddy to make a 16:9 image of a lighthouse at dusk in watercolor.
Cursor
- Get your AIBuddy API key. It starts with ab_live_. Create a key
-
Paste this into ~/.cursor/mcp.json, replacing ab_live_... with your key:
{ "mcpServers": { "aibuddy": { "command": "npx", "args": ["-y", "github:ThomasWDev/aibuddy-mcp"], "env": { "AIBUDDY_API_KEY": "ab_live_..." } } } } - Restart the client, then try:
Use aibuddy to make a 16:9 image of a lighthouse at dusk in watercolor.
VS Code
- Get your AIBuddy API key. It starts with ab_live_. Create a key
-
Paste this into .vscode/mcp.json in your project, replacing ab_live_... with your key. Keep that file out of version control while it holds a key.
{ "servers": { "aibuddy": { "type": "stdio", "command": "npx", "args": ["-y", "github:ThomasWDev/aibuddy-mcp"], "env": { "AIBUDDY_API_KEY": "ab_live_..." } } } } - Restart the client, then try:
Use aibuddy to make a 16:9 image of a lighthouse at dusk in watercolor.
Codex
- Get your AIBuddy API key. It starts with ab_live_. Create a key
-
Add this to ~/.codex/config.toml, replacing ab_live_... with your key:
[mcp_servers.aibuddy] command = "npx" args = ["-y", "github:ThomasWDev/aibuddy-mcp"] env = { AIBUDDY_API_KEY = "ab_live_..." } - Restart the client, then try:
Use aibuddy to make a 16:9 image of a lighthouse at dusk in watercolor.
Command line
- Get your AIBuddy API key. It starts with ab_live_. Create a key
-
Export your key, then call any tool straight from the shell. Add --json to any command for machine-readable output.
export AIBUDDY_API_KEY=ab_live_... npx -y github:ThomasWDev/aibuddy-mcp models npx -y github:ThomasWDev/aibuddy-mcp image "a lighthouse at dusk, watercolor" --aspect 16:9 --out ./art npx -y github:ThomasWDev/aibuddy-mcp edit room.jpg "the same room, mid-century modern" npx -y github:ThomasWDev/aibuddy-mcp video "waves rolling onto a beach at sunrise" --image still.jpg --wait 300 npx -y github:ThomasWDev/aibuddy-mcp music "upbeat synthwave for a product demo" --instrumental --wait 300 npx -y github:ThomasWDev/aibuddy-mcp task <task_id> --json - Start with the free model list:
npx -y github:ThomasWDev/aibuddy-mcp models
Just describe it
Your agent picks the right tool from a plain-language request. Six tools, one key.
edit_image
A restyled variant of your image. Works on a JPEG of about 1024px, up to 480 KB.
Try: Restyle room.jpg as a bright Scandinavian living room.
generate_video
A short clip from text or a still image. Returns a task id at once, or waits up to wait_seconds for the result.
Try: Turn still.jpg into a short clip of waves rolling in at sunrise.
generate_music
A song with vocals or an instrumental from a text prompt. Runs in the background and returns a task id.
Try: Write an upbeat instrumental for a product demo video.
get_task
Status and result URLs for a video or music task. Task ids expire after 24 hours.
Try: Is my video ready yet? Give me the link.
list_models
The image, video and music models available right now. No key needed.
Try: Which AIBuddy video models can I use?
Models available today
The list changes as models are added. list_models always returns the current set.
Image
Video
Music
Frequently asked questions
How am I billed?
Each generation spends credits from the AIBuddy key in your config, the same credits the AIBuddy API uses. A new key includes a free trial of $1.00 USD of API usage (100 Credits) with no credit card. After that you buy pay-as-you-go credits, and credits never expire.
Do I need keys from other AI providers?
No. Your AIBuddy key covers every model on this page. AIBuddy calls the model providers for you.
Where do my files go?
Images come back inline in the chat. Set AIBUDDY_OUTPUT_DIR in the env block to also save them to a folder. Video and music run as background tasks: get_task returns the result URLs, and task ids expire after 24 hours, so download anything you want to keep. On the command line, --out picks the folder.
Which MCP versions are supported?
The server speaks MCP protocol version 2026-07-28 and also accepts the older initialize handshake, so clients on either version connect.
Is my key safe?
Your key stays in your local client config and is sent only to the AIBuddy API. Treat it like a password and keep it out of shared repositories.
Ready to connect your agent?
Create a free key on the AIBuddy API portal, paste one config, and start generating.
Get a free API key Back to the AIBuddy API portal