MCP 服务器

dida365-agent

MCP server that enables AI agents to manage Dida365/TickTick tasks, projects, tags, and habits through natural language, with full CRUD, search, and advanced V2 capabilities.

README

<h1 align="center">dida365-agent</h1>

Let AI agents manage your Dida365 / TickTick — CLI · Skill · MCP

English | <a href="README_zh.md">中文</a>

Let AI agents manage Dida365 / TickTick tasks, projects, tags, and habits through natural language. Three form factors: a lightweight CLI, a ready-to-install Agent Skill, and a standard MCP Server.

No install required — run it directly with uvx. Supports both Dida365 (China) and TickTick (International), switchable with one env var.

Features

<table> <tr><td>Full task management</td><td>Create, update, complete, delete, and move tasks with priority, tags, dates, reminders, recurrence, and checklist items — plus batch operations.</td></tr> <tr><td>Projects & queries</td><td>CRUD projects (list / kanban / timeline views), filter by project, date, priority, tag, and status, and review completed tasks.</td></tr> <tr><td>V2 advanced power</td><td>Server-side full-text search, tag management, habit check-ins, project folders, parent/child tasks, task pinning — capabilities the official Open API doesn't cover.</td></tr> <tr><td>Three form factors</td><td>CLI (run-and-exit, saves resources and context), Agent Skill (one-command install with built-in methodology), MCP Server (long-running, tool calls).</td></tr> <tr><td>Dual platform</td><td>Switch between Dida365 (China) and TickTick (International) with <code>DIDA365_REGION</code>.</td></tr> <tr><td>Agent-friendly</td><td>Structured JSON to stdout by default, errors to stderr with non-zero exit codes — easy for agents to parse and orchestrate.</td></tr> </table>

Quick Start

Option 1: Use the CLI directly

No install needed — run with uvx (requires uv). First prepare credentials per the configuration guide and write them to .env:

# Browser OAuth (saves token, ~180 days)
uvx dida365-agent dida auth login

# List all projects
uvx dida365-agent dida project list

# Create a high-priority task due tomorrow
uvx dida365-agent dida task create --title "Review PR" --project <projectId> \
  --priority 5 --due-date "2026-05-30T18:00:00+0800"

# Full-text search (V2)
uvx dida365-agent dida search "meeting"

After installing locally (uv tool install dida365-agent), use the shorter dida command.

Option 2: Use as an Agent Skill

Install the Skill in any Skill-compatible AI tool (Claude Code, Cursor, etc.) and drive it with natural language:

# Install the Skill
npx skills add linhai0872/dida365-agent

Then just describe what you need in the AI chat:

Tidy up my unfinished tasks for today, list the high-priority ones, and move the overdue ones to the "Later" project

The Skill recognizes intent, fills in missing details, and assembles dida commands automatically — no manual parameters required.

Option 3: As an MCP Server

Exposes 44 tools as a standard MCP Server for Claude Code, Cursor, Windsurf, etc. See MCP Server integration.

Documentation

Doc	Contents
CLI Reference	All commands, parameters, conventions
Configuration	Credentials, enabling V2, env vars, token lifecycle
MCP Server	Local / source / Docker deployment, AI tool config

Development

uv sync                              # Install dependencies
uv run python -m pytest tests/       # Run tests
uv run ruff check src/ tests/        # Lint
uv run dida --help                   # Run the CLI locally