MCP DeepInfra AI Tools Server

Instructions

Implementation Reference

• src/mcp_deepinfra/server.py:85-96 (handler)

  The core handler function that executes the embeddings tool logic, generating embeddings for a list of input texts using the DeepInfra OpenAI-compatible API.

  async def embeddings(inputs: list[str]) -> str:
      """Generate embeddings for a list of texts using DeepInfra OpenAI-compatible API."""
      model = DEFAULT_MODELS["embeddings"]
      try:
          response = await client.embeddings.create(
              model=model,
              input=inputs,
          )
          embeddings_list = [item.embedding for item in response.data]
          return str(embeddings_list)
      except Exception as e:
          return f"Error generating embeddings: {type(e).__name__}: {str(e)}"

• src/mcp_deepinfra/server.py:83-84 (registration)

  Conditional registration of the embeddings tool using FastMCP's @app.tool() decorator based on ENABLED_TOOLS configuration.

  if "all" in ENABLED_TOOLS or "embeddings" in ENABLED_TOOLS:
      @app.tool()

• src/mcp_deepinfra/server.py:85-85 (schema)

  Function signature providing the input schema (inputs: list[str]) and output type (str) for automatic tool schema generation in FastMCP.

  async def embeddings(inputs: list[str]) -> str:

• src/mcp_deepinfra/server.py:34-34 (helper)

  Configuration for the default model used by the embeddings tool.

  "embeddings": os.getenv("MODEL_EMBEDDINGS", "sentence-transformers/all-MiniLM-L6-v2"),