Get笔记 MCP Server

Instructions

Implementation Reference

• src/client.ts:71-104 (handler)

  Core handler function that implements the knowledge_search tool logic: applies rate limiting, handles default topic_ids, makes HTTP POST to '/knowledge/search' API, and processes errors.

  async knowledgeSearch(params: KnowledgeSearchRequest): Promise<any> {
    await this.rateLimiter.waitForSlot();
  
    // 如果没有提供topic_ids且配置了默认topic_id,则使用默认值
    if (!params.topic_ids || params.topic_ids.length === 0) {
      if (this.config.defaultTopicId) {
        params.topic_ids = [this.config.defaultTopicId];
        logger.debug('Using default topic_id', { topic_id: this.config.defaultTopicId });
      } else {
        throw new Error('topic_ids is required or set GET_BIJI_DEFAULT_TOPIC_ID in environment');
      }
    }
  
    try {
      const response = await this.client.post('/knowledge/search', params);
      return response.data;
    } catch (error: any) {
      const errorMsg = error.response?.data?.h?.e || error.message;
      const statusCode = error.response?.status;
      
      logger.error('Knowledge search failed', { 
        error: errorMsg, 
        statusCode,
        params: { ...params, topic_ids: params.topic_ids },
        responseData: error.response?.data 
      });
      
      if (statusCode === 403) {
        throw new Error(`Knowledge search failed: Authentication error (403). Please check your GET_BIJI_API_KEY is valid.`);
      }
      
      throw new Error(`Knowledge search failed: ${errorMsg}`);
    }
  }

• src/index.ts:152-177 (handler)

  MCP server tool execution handler for 'knowledge_search': parses arguments, calls client.knowledgeSearch, and returns JSON-formatted result as text content.

  case 'knowledge_search': {
    const { question, topic_ids, deep_seek, refs, history } = args as {
      question: string;
      topic_ids?: string[];
      deep_seek: boolean;
      refs?: boolean;
      history?: ChatMessage[];
    };
  
    const result = await client.knowledgeSearch({
      question,
      topic_ids,
      deep_seek,
      refs,
      history,
    });
  
    return {
      content: [
        {
          type: 'text',
          text: JSON.stringify(result, null, 2),
        },
      ],
    };
  }

• src/index.ts:38-78 (registration)

  Tool registration in MCP server's tools list: defines name 'knowledge_search', description, and input schema for validation.

  {
    name: 'knowledge_search',
    description: '在Get笔记知识库中进行AI增强搜索,返回经过深度处理的答案。支持对话历史追问。',
    inputSchema: {
      type: 'object',
      properties: {
        question: {
          type: 'string',
          description: '要搜索的问题',
        },
        topic_ids: {
          type: 'array',
          items: { type: 'string' },
          description: '知识库ID列表(当前只支持1个)。如果配置了GET_BIJI_DEFAULT_TOPIC_ID环境变量,可省略此参数',
        },
        deep_seek: {
          type: 'boolean',
          description: '是否启用深度思考',
          default: true,
        },
        refs: {
          type: 'boolean',
          description: '是否返回引用来源',
          default: false,
        },
        history: {
          type: 'array',
          items: {
            type: 'object',
            properties: {
              content: { type: 'string' },
              role: { type: 'string', enum: ['user', 'assistant'] },
            },
            required: ['content', 'role'],
          },
          description: '对话历史,用于追问场景',
        },
      },
      required: ['question', 'deep_seek'],
    },
  },

• src/types.ts:6-12 (schema)

  TypeScript type definition for KnowledgeSearchRequest interface, matching the tool's input schema, used in client code.

  export interface KnowledgeSearchRequest {
    question: string;
    topic_ids?: string[];  // 可选,如果配置了defaultTopicId
    deep_seek: boolean;
    refs?: boolean;
    history?: ChatMessage[];
  }