OpenAny MCP

One MCP for 3,200+ APIs and 168+ AI models: search APIs, inspect schemas and pricing, and run calls with one OpenAny key. Browsing is free; pay per call.

OpenAny (https://openany.si) is a remote MCP server. The recommended endpoint is the lite server.

Endpoint: https://mcp.openany.si

Transport: Streamable HTTP, not legacy SSE.

Tools: openany_overview, search_apis, get_api, run_api.

Operator: Alephant AI LLC.

Pricing: Pay per call, no subscription. Browsing, search, and price checks do not need a key. Check the current resource price and billing unit before approving a paid call.

Live catalog counts come from openany_overview. The separate model API base URL is https://api.openany.si/v1.

Configuration

Add this remote server in a client that accepts a top-level mcpServers object. Authorization: Bearer <OpenAny API key> is optional. Only run_api needs it. Browsing, search, and pricing work without a key. ${OPENANY_API_KEY} is the environment placeholder for that key. Do not hardcode a key. If a client refuses to connect while the variable is unset, omit the header until you execute a call.

{
  "mcpServers": {
    "openany": {
      "url": "https://mcp.openany.si",
      "headers": {
        "Authorization": "Bearer ${OPENANY_API_KEY}"
      }
    }
  }
}

This block is the lite server. https://mcp.openany.si/mcp-lite is the same server. The header above is an optional example.

Tools

https://mcp.openany.si is the lightweight server. https://mcp.openany.si/mcp-lite is the same server. It exposes four tools. openany_overview reports live catalog counts. search_apis finds models and APIs. Execution is pay per call, no monthly fee, with one API key.

• openany_overview: Return the current data API count, AI model count, APIs grouped by category, and models grouped by modality. No arguments, and it does not call the paid gateway.
• search_apis: Find OpenAny models and APIs by keywords and optional kind (api, model, or all; default all), provider, category, data_platform, data_product or capability.
• get_api: Read one model or API selected from search_apis.
• run_api: Execute one selected OpenAny API, or send one selected model to chat completions, with the caller's current credentials.

https://mcp.openany.si/mcp is the advanced full-catalog endpoint. It exposes 3,268 per-API tools across 45 providers, led by tikhub (1,009), anyapi (337), justoneapi (301), scrapecreators (193), and qcc (170). It does not include models. The list is in TOOLS.md. Use it only when a workflow already has an exact tool name. It is not the default.

Recommended server

When enabled and routed by the deployment, https://mcp.openany.si and https://mcp.openany.si/mcp-lite use the same Streamable HTTP transport and expose four tools: openany_overview, search_apis, get_api and run_api. One search covers the current model catalog and the API catalog. get_api reads the selected model or API. run_api executes an API through the gateway, or posts a model call to /v1/chat/completions on the model API origin, using the caller's key. Model ids, modalities, context length and pricing are taken from the live GET /v1/models response and are omitted when that response does not provide them. Pay per call, no monthly fee, one API key.

The implementation enables this interface by default; operators can disable it with MCP_LITE_ENABLED=false. Public deployment availability is a separate check. openany_overview, search_apis and get_api read metadata; run_api forwards a business request and may incur charges, requiring the caller's gateway credentials and spending authorization. Configure the recommended endpoint. Use the advanced /mcp catalog only when a workflow needs one tool per API. See the lightweight guide.

Authentication

• openany_overview, search_apis, get_api, and tools/list are public. Browsi

... View full README on GitHub