MCP Backlog Server

import { tool } from "@opencode-ai/plugin"; import { readTodos, updateTodoStatus, validateDependencies, listTodos } from "../lib/backlog-todo-shared"; export default tool({ description: "Mark backlog todos as complete with dependency validation", args: { action: tool.schema .enum(["done", "list"]) .describe("Action to perform"), topic: tool.schema .string() .describe("Backlog topic"), todoId: tool.schema .string() .optional() .describe("Todo ID to mark complete (required for done)"), status: tool.schema .enum(["pending", "in_progress", "completed", "cancelled"]) .optional() .describe("Filter for list action"), batch: tool.schema .string() .optional() .describe("Filter for list action"), }, async execute(args, context) { const { action, topic, todoId, status, batch } = args; if (action === "done") { if (!todoId) { throw new Error("todoId is required for done action"); } const data = await readTodos(topic); // Validate dependencies const validation = validateDependencies(data.todos, todoId); if (!validation.valid) { const errors = []; if (validation.missing.length > 0) { errors.push(`Missing dependencies: ${validation.missing.join(", ")}`); } if (validation.incomplete.length > 0) { errors.push(`Incomplete dependencies: ${validation.incomplete.join(", ")}`); } throw new Error(`Cannot mark todo ${todoId} as complete. ${errors.join(". ")}`); } // Mark complete const todo = await updateTodoStatus(topic, todoId, "completed"); return JSON.stringify({ completed: todo, hint: "Todo completed. Continue remaining todos or submit backlog for review" }, null, 2); } if (action === "list") { const todos = await listTodos(topic, { status, batch }); if (todos.length === 0) { return `No todos found for backlog: ${topic}`; } return JSON.stringify({ backlogTopic: topic, todos, hint: "Review completed todos and continue remaining work" }, null, 2); } throw new Error(`Unknown action: ${action}`); } });