MCP DeepInfra AI Tools Server

Instructions

Implementation Reference

• src/mcp_deepinfra/server.py:66-81 (handler)

  The handler function that implements the core logic of the text_generation tool, using AsyncOpenAI client to call DeepInfra's completions API with configurable model, fixed max_tokens=256 and temperature=0.7.

  async def text_generation(prompt: str) -> str:
      """Generate text completion using DeepInfra OpenAI-compatible API."""
      model = DEFAULT_MODELS["text_generation"]
      try:
          response = await client.completions.create(
              model=model,
              prompt=prompt,
              max_tokens=256,
              temperature=0.7,
          )
          if response.choices:
              return response.choices[0].text
          else:
              return "No text generated"
      except Exception as e:
          return f"Error generating text: {type(e).__name__}: {str(e)}"

• src/mcp_deepinfra/server.py:64-65 (registration)

  The conditional registration of the text_generation tool via FastMCP's @app.tool() decorator, enabled if 'all' or 'text_generation' is in ENABLED_TOOLS.

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

• src/mcp_deepinfra/server.py:33-33 (helper)

  Helper configuration defining the default model ID for the text_generation tool, overridable via MODEL_TEXT_GENERATION env var.

  "text_generation": os.getenv("MODEL_TEXT_GENERATION", "meta-llama/Llama-2-7b-chat-hf"),

• src/mcp_deepinfra/server.py:66-66 (schema)

  Inferred schema from function signature: input 'prompt' as str, output str. Docstring provides description for MCP tool schema.

  async def text_generation(prompt: str) -> str: