ai_summarize

Instructions

Implementation Reference

• src/onion_mcp_server/tools/ai.py:85-114 (schema)

  Tool schema registration for ai_summarize with inputSchema defining required 'text', optional 'mode' (bullets/paragraph/one_line), 'max_points' (integer), and 'language' fields.

  types.Tool(
      name="ai_summarize",
      description="对长文本进行摘要，支持要点列表、段落摘要、一句话摘要三种模式。",
      inputSchema={
          "type": "object",
          "properties": {
              "text": {
                  "type":        "string",
                  "description": "要摘要的文本",
              },
              "mode": {
                  "type":        "string",
                  "description": "摘要模式: bullets（要点列表）/ paragraph（段落）/ one_line（一句话）",
                  "enum":        ["bullets", "paragraph", "one_line"],
                  "default":     "bullets",
              },
              "max_points": {
                  "type":        "integer",
                  "description": "要点数量（bullets 模式有效，默认 5）",
                  "default":     5,
              },
              "language": {
                  "type":        "string",
                  "description": "输出语言（默认与原文相同）",
                  "default":     "",
              },
          },
          "required": ["text"],
      },
  ),

• src/onion_mcp_server/tools/ai.py:234-251 (handler)

  The handler function (handle_ai) that executes ai_summarize logic: reads mode/max_points/language params, constructs LLM prompt with appropriate instruction, calls llm_call, and returns TextContent.

  elif name == "ai_summarize":
      mode       = a.get("mode", "bullets")
      max_points = int(a.get("max_points", 5))
      lang_str   = f"，用{a['language']}输出" if a.get("language") else ""
  
      if mode == "bullets":
          instruction = f"提炼出最重要的 {max_points} 个要点，用 Markdown 列表格式输出{lang_str}"
      elif mode == "paragraph":
          instruction = f"写成一段连贯的摘要段落{lang_str}"
      else:  # one_line
          instruction = f"用一句话概括核心内容{lang_str}"
  
      prompt = (
          f"请对以下文本进行摘要，{instruction}。\n\n"
          f"---\n{a['text']}\n---"
      )
      reply = await llm_call(prompt)
      return [types.TextContent(type="text", text=reply)]

• src/onion_mcp_server/server.py:48-51 (registration)

  Server registration: ai_summarize (as part of AI_TOOLS) is mapped to the handle_ai handler via the _HANDLERS routing table.

  # ── 路由表 ────────────────────────────────────────────────────
  _HANDLERS: dict = {}
  for _t in AI_TOOLS:     
      _HANDLERS[_t.name] = handle_ai

• src/onion_mcp_server/tools/_llm.py:78-88 (helper)

  The llm_call helper function called by the ai_summarize handler to make the actual LLM API request.

  async def llm_call(
      prompt: str,
      system: Optional[str] = None,
      temperature: float = 0.7,
  ) -> str:
      """单轮调用"""
      messages = []
      if system:
          messages.append({"role": "system", "content": system})
      messages.append({"role": "user", "content": prompt})
      return await llm_chat(messages, temperature=temperature)