LaunchNotes MCP Server

Instructions

Implementation Reference

• src/feedback/tools.ts:63-117 (handler)

  The asynchronous handler function that implements the core logic of the 'launchnotes_search_feedback' tool. It performs the GraphQL search query via searchFeedback, processes the results, and returns formatted JSON or Markdown output based on parameters. Handles errors gracefully.

  async (params: SearchFeedbackInput) => {
    try {
      const result = await searchFeedback(client, {
        projectId: params.project_id,
        query: params.query,
        reaction: params.reaction,
        importance: params.importance,
        organizedState: params.organized_state,
        starred: params.starred,
        archived: params.archived,
        first: params.limit,
      });
  
      const feedbacks = result.project.feedbacks.nodes;
  
      if (params.response_format === RESPONSE_FORMAT.JSON) {
        return {
          content: [
            {
              type: "text",
              text: JSON.stringify(
                {
                  total: feedbacks.length,
                  feedbacks,
                  hasMore: result.project.feedbacks.pageInfo.hasNextPage,
                },
                null,
                2
              ),
            },
          ],
        };
      }
  
      // Markdown format
      return {
        content: [
          {
            type: "text",
            text: formatFeedbackListMarkdown(feedbacks),
          },
        ],
      };
    } catch (error) {
      return {
        isError: true,
        content: [
          {
            type: "text",
            text: `Error searching feedback: ${error instanceof Error ? error.message : "Unknown error"}`,
          },
        ],
      };
    }
  }

• src/feedback/schemas.ts:35-68 (schema)

  Zod schema defining the input parameters and validation for the 'launchnotes_search_feedback' tool, including project_id, filters like query, reaction, importance, and output format.

  export const SearchFeedbackSchema = z
    .object({
      project_id: z
        .string()
        .min(1, "Project ID is required")
        .describe("The ID of the LaunchNotes project"),
      query: z
        .string()
        .optional()
        .describe("Search term to find in feedback content"),
      reaction: reactionSchema,
      importance: importanceSchema,
      organized_state: z
        .string()
        .optional()
        .describe("Filter by organized state: 'organized', 'unorganized', 'announcement', 'idea', 'roadmap'"),
      starred: z
        .boolean()
        .optional()
        .describe("Filter by starred status"),
      archived: z
        .boolean()
        .optional()
        .describe("Filter by archived status"),
      limit: z
        .number()
        .min(1)
        .max(100)
        .default(20)
        .optional()
        .describe("Number of feedback items to return (max 100)"),
      response_format: responseFormatSchema,
    })
    .strict();

• src/feedback/tools.ts:26-62 (registration)

  The server.registerTool call that registers the 'launchnotes_search_feedback' tool, including its name, title, detailed description, input schema reference, and annotations.

      "launchnotes_search_feedback",
      {
        title: "Search LaunchNotes Feedback",
        description: `Search and filter customer feedback in a LaunchNotes project.
  
  Args:
    - project_id (string): The ID of the project (required)
    - query (string, optional): Search term to find in feedback content
    - reaction ('happy' | 'meh' | 'sad', optional): Filter by customer sentiment
    - importance ('low' | 'medium' | 'high', optional): Filter by importance level
    - organized_state (string, optional): Filter by state ('organized', 'unorganized', 'announcement', 'idea', 'roadmap')
    - starred (boolean, optional): Filter by starred status
    - archived (boolean, optional): Filter by archived status
    - limit (number, optional): Number to return (max 100, default: 20)
    - response_format ('json' | 'markdown'): Output format (default: 'markdown')
  
  Returns:
    List of feedback items with content, sentiment, importance, customer info, and timestamps
  
  Use Cases:
    - "What are customers saying about Digests?"
    - "Show me all unhappy feedback"
    - "Find high importance feedback that's unorganized"
    - "Search feedback containing 'API integration'"
    - "Show me starred feedback"
  
  Error Handling:
    - Returns "Project not found" if project ID doesn't exist
    - Returns "Authentication failed" if API token is invalid`,
        inputSchema: SearchFeedbackSchema,
        annotations: {
          readOnlyHint: true,
          destructiveHint: false,
          idempotentHint: true,
          openWorldHint: true,
        },
      },