apiany mcp

What This Repository Provides

APIAny MCP gives AI agents a structured way to understand and use APIAny:

• Discover public APIAny models by provider, modality, capability, and pricing metadata.
• Estimate APIAny credits before choosing a model or running a workflow.
• Fetch compact documentation context from APIAny docs and llms.txt.
• Generate OpenAI-compatible cURL, Python, JavaScript, Go, Java, and PHP examples.
• Create image or video generation tasks only when an API key is configured and the user explicitly confirms paid usage.
• Package an agent skill at skills/apiany-integration so supported agents can learn the APIAny workflow.

Quick Start

Run the server directly with npm:

npx -y @apiany-ai/mcp

Or install and run it from this repository:

git clone git@github.com:ailingqu/ApiAny.AI-MCP.git
cd ApiAny.AI-MCP
npm install
npm start

Visual Setup

1. Install and Start the MCP Server

2. Add APIAny MCP to Your Agent Client

Use this configuration in Claude Code, Cursor, VS Code, or another MCP-compatible client:

{
  "mcpServers": {
    "apiany": {
      "command": "npx",
      "args": ["-y", "@apiany-ai/mcp"],
      "env": {
        "APIANY_BASE_URL": "https://apiany.ai"
      }
    }
  }
}

For paid image or video task tools, also set APIANY_API_KEY:

{
  "APIANY_BASE_URL": "https://apiany.ai",
  "APIANY_API_KEY": "your_apiany_api_key"
}

3. Ask Your Agent to Use APIAny

Example prompts:

Use APIAny MCP to find an image model for product photos, compare pricing, and show JavaScript integration code.

Use APIAny MCP to estimate credits for 20 gpt-style chat requests with 10k input tokens and 5k output tokens.

Use APIAny MCP to show JavaScript usage for nano-banana-pro and Python usage for veo3-1-fast.

MCP Tools

Tool	Purpose	API key
list_models	List public APIAny models with pricing and capability metadata.	No
search_models	Search models by text, provider, modality, or capability.	No
get_model	Get one public model by model id or display name.	No
estimate_cost	Estimate credits from public pricing metadata.	No
get_model_usage	Return endpoint, payload, async behavior, and language examples for models.	No
get_integration_examples	Return cURL, Python, JavaScript, Go, Java, or PHP examples.	No
get_docs_context	Return compact APIAny documentation context from /llms.txt.	No
create_image_task	Create a paid async image task after explicit confirmation.	Yes
create_video_task	Create a paid async video task after explicit confirmation.	Yes
get_task_status	Read async media task status.	Yes

Agent Skill

The APIAny Integration skill lives in skills/apiany-integration.

It helps agents:

• Choose APIAny models by modality, provider, pricing, and capability.
• Generate examples for chat, image, video, and media workflows.
• Explain async task creation and polling.
• Require explicit confirmation before paid generation tasks.

Environment Variables

Variable	Required	Description
APIANY_BASE_URL	No	APIAny base URL. Defaults to https://apiany.ai.
APIANY_API_KEY	Only for paid tools	API key used for paid image/video task creation and task polling.

Safety Model

Read-only tools work without an API key. Paid generation tools require both:

• APIANY_API_KEY
• confirm_paid_request=true

The server does not persist API keys. It only reads them from the current process environment.

Development

npm install
npm run check
npx -y @modelcontextprotocol/inspector node src/server.js

Useful files:

• src/server.js: MCP server implementation.
• examples/mcp.json: client configuration example.
• docs/distribution.md: release checklist for npm, MCP directories, and skill marketplaces.
• CHANGELOG.md: release history.

Links

• Website: https://apiany.ai
• GitHub: https://github.com/ailingqu/ApiAny.AI-MCP
• npm: https://www.npmjs.com/package/@apiany-ai/mcp
• Models: https://apiany.ai/models
• Docs context: https://apiany.ai/llms.txt

License

MIT