Skip to main content
Glama

recall_memory_tool

Retrieve relevant information from project memory using keywords, questions, or statements to support decision-making and maintain context.

Instructions

从项目记忆中回忆相关信息

Args:
    interest: 想要回忆的任何东西,可以是一句陈述、一个问题甚至是关键词

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault
interestYes

Implementation Reference

  • The primary handler function for the 'recall_memory_tool' MCP tool. It is registered via @mcp.tool() decorator and delegates the core logic to recall_memory.
    @mcp.tool()
    async def recall_memory_tool(interest: str) -> str:
        """从项目记忆中回忆相关信息
    
        Args:
            interest: 想要回忆的任何东西,可以是一句陈述、一个问题甚至是关键词
        """
        return await recall_memory(interest, _registry)
  • Helper function implementing the recall logic using a small LLM agent with ListMemoriesTool, ReadMemoryTool, and a submit tool for generating the final report.
    async def recall_memory(interest: str, registry: MemoryRegistry) -> str:
        """使用 small_agent 实现的回忆流程"""
        list_tool = ListMemoriesTool(registry)
        read_tool = ReadMemoryTool(registry)
    
        final_tools: list[ToolUnionParam] = [
            {
                "name": "submit",
                "description": "提交回忆报告",
                "input_schema": {
                    "type": "object",
                    "properties": {
                        "report": {
                            "type": "string",
                            "description": "综合回忆报告(Markdown 格式)",
                        }
                    },
                    "required": ["report"],
                },
            }
        ]
    
        initial_prompt = f"""尝试从记忆库回忆与这个有关的信息:{interest}
    
    请按以下指导处理:
    1. 使用 list_memories 工具搜索相关记忆
    2. 使用 read_memory 工具读取相关记忆
    3. 基于读取的内容,调用 submit 提交综合报告
    
    提示:
    - list_memories 返回的是按匹配度排序的结果
    - read_memory 需要提供每个记忆唯一的关键词组
    - 报告除了直接回应用户的兴趣点,还应该包含补充详细的背景信息
    - 如果没有相关内容,请报告'没有相关记忆'"""
    
        result = await small_agent(
            initial_prompt=initial_prompt,
            tools=[list_tool, read_tool],
            final=final_tools,
            maxIter=10,
        )
    
        if result is None:
            return "查询超时,未能生成报告"
    
        tool_name, tool_input = result
        if tool_name == "submit":
            return tool_input.get("report", "")
    
        return "未知错误"
Install Server

Other Tools

Latest Blog Posts

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/JerryZhongJ/memory-mcp'

If you have feedback or need assistance with the MCP directory API, please join our Discord server