dida365

滴答清单 / dida365（TickTick）官方 MCP 服务 的 Python SDK。把 32 个 MCP 工具包装成同名方法，同时提供同步与异步接口、完整类型标注、自动重试与凭据隔离。

A Python SDK for the official dida365 / TickTick MCP server. Wraps all 32 MCP tools as Python methods with sync + async clients, full type stubs, automatic retries, and safe credential handling.

中文文档

安装

pip install dida365-sdk            # 仅同步客户端
pip install "dida365-sdk[async]"   # 同时安装 httpx，启用 AsyncClient

安装名是 dida365-sdk，导入名仍是 import dida365。

获取 Token

打开 滴答清单网页版，进入 头像 → 设置 → 账户与安全 → API 口令，新建并复制 token。

设置环境变量（推荐）：

export DIDA365_TOKEN=dp_xxxxxxxxxxxxxxxxxxxxxxxxxx

或者在项目目录创建 .env 文件：

DIDA365_TOKEN=dp_xxxxxxxxxxxxxxxxxxxxxxxxxx

快速开始

import dida365

with dida365.Client() as client:
    # 1. 列出所有清单
    projects = client.list_projects()

    # 2. 创建任务
    task = client.create_task(task={
        "projectId": projects[0]["id"],
        "title": "明天提交周报",
        "priority": 5,                    # 0=无 1=低 3=中 5=高
        "startDate": "2026-05-09T09:00:00+08:00",
        "isAllDay": False,
        "timeZone": "Asia/Shanghai",
        "tags": ["work"],
    })

    # 3. 查询今日待办
    today = client.list_undone_tasks_by_time_query(query_command="today")

    # 4. 完成任务
    client.complete_task(project_id=task["projectId"], task_id=task["id"])

异步用法

import asyncio
import dida365

async def main() -> None:
    async with dida365.AsyncClient() as client:
        projects, today = await asyncio.gather(
            client.list_projects(),
            client.list_undone_tasks_by_time_query(query_command="today"),
        )
        print(f"{len(projects)} projects, {len(today)} tasks today")

asyncio.run(main())

错误处理

import dida365

try:
    client = dida365.Client(token="bad-token")
except dida365.AuthenticationError as e:
    print(f"token 失效: {e.code}")          # 401 / 403
except dida365.MCPProtocolError as e:
    print(f"网络/协议错误: {e}")
except dida365.ToolError as e:
    print(f"工具调用失败: {e.tool} -> {e}")

全部 32 个工具

每个工具的参数与返回类型已经在 dida365/client.pyi 中声明，IDE / mypy / pyright 都能直接获取补全。

高级配置

client = dida365.Client(
    token="dp_xxx",                  # 显式 token（默认从 env 读取）
    url="https://mcp.dida365.com",   # MCP 服务地址
    timeout=(5.0, 30.0),             # (connect, read)
    max_retries=3,                   # 5xx / 429 / 网络错误的重试次数
)

Client 与 AsyncClient 都自动遵循 Retry-After 头部、自动透传 Mcp-Session-Id 与 MCP-Protocol-Version 头部。

安全约定

• repr(client) 不会泄露 token；只显示 Client(url=...)。
• 进程环境变量只接受 DIDA365_TOKEN，避免与其他工具的 TOKEN 冲突。
• .env 中可使用 DIDA365_TOKEN 或 TOKEN（仅在 cwd/.env 中生效）。

开发

本仓库使用 uv 管理 Python 环境与依赖。

git clone https://github.com/shawn-bluce/dida365-sdk.git
cd dida365-sdk
uv sync --all-extras           # 创建 .venv 并安装 dev + async 依赖

# 单元测试（不联网）
uv run pytest

# 实测脚本（会创建并关闭一个沙箱清单）
DIDA365_TOKEN=dp_xxx uv run python tests/integration/test_smoke_live.py

# 重新生成类型存根
uv run python scripts/generate_stubs.py --refresh

# 打包与发布
uv build                       # 生成 dist/*.whl 与 dist/*.tar.gz
uv publish                     # 上传到 PyPI（需配置 token）

English Documentation

Install

pip install dida365-sdk            # sync client only
pip install "dida365-sdk[async]"   # also installs httpx for AsyncClient

The distribution name is dida365-sdk; the import name remains import dida365.

Get a token

In the dida365 web app, open Profile → Settings → Account & Security → API Token, create and copy a new token. Then either:

export DIDA365_TOKEN=dp_xxxxxxxxxxxxxxxxxxxxxxxxxx

…or drop a .env file at the project root with DIDA365_TOKEN=....

Quickstart

import dida365

with dida365.Client() as client:
    projects = client.list_projects()

    task = client.create_task(task={
        "projectId": projects[0]["id"],
        "title": "Submit weekly report",
        "priority": 5,
        "startDate": "2026-05-09T09:00:00+08:00",
        "isAllDay": False,
        "timeZone": "Asia/Shanghai",
        "tags": ["work"],
    })

    today = client.list_undone_tasks_by_time_query(query_command="today")
    client.complete_task(project_id=task["projectId"], task_id=task["id"])

Async usage

import asyncio
import dida365

async def main() -> None:
    async with dida365.AsyncClient() as client:
        projects, today = await asyncio.gather(
            client.list_projects(),
            client.list_undone_tasks_by_time_query(query_command="today"),
        )
        print(f"{len(projects)} projects, {len(today)} tasks today")

asyncio.run(main())

Error handling

import dida365

try:
    client = dida365.Client(token="bad-token")
except dida365.AuthenticationError as e:
    ...                                  # HTTP 401 / 403
except dida365.MCPProtocolError as e:
    ...                                  # network / HTTP / JSON-RPC error
except dida365.ToolError as e:
    ...                                  # tool reported isError=true

Tool catalogue

All 32 server-advertised tools are dispatched dynamically by name and statically typed via client.pyi / async_client.pyi. Categories: projects (5), tasks (9), querying (7), habits (7), focus (3), preferences (1).

Advanced configuration

client = dida365.Client(
    token="dp_xxx",                   # explicit token (defaults to env)
    url="https://mcp.dida365.com",    # MCP endpoint
    timeout=(5.0, 30.0),              # (connect, read)
    max_retries=3,                    # 5xx / 429 / network retries
)

Both Client and AsyncClient honor Retry-After, round-trip Mcp-Session-Id, and send MCP-Protocol-Version on every non-initialize request.

Security defaults

• repr(client) masks the bearer token.
• Only DIDA365_TOKEN is read from process env (avoids collisions with TOKEN set by other tools).
• .env in the current working directory accepts DIDA365_TOKEN or TOKEN.

Development

This repo uses uv to manage the Python environment and dependencies.

git clone https://github.com/shawn-bluce/dida365-sdk.git
cd dida365-sdk
uv sync --all-extras           # creates .venv and installs dev + async extras

# Unit tests (offline)
uv run pytest

# Live integration smoke test (creates a sandbox project, closes it after)
DIDA365_TOKEN=dp_xxx uv run python tests/integration/test_smoke_live.py

# Regenerate type stubs from the live MCP catalogue
uv run python scripts/generate_stubs.py --refresh

# Build & publish
uv build                       # produces dist/*.whl and dist/*.tar.gz
uv publish                     # uploads to PyPI (requires token)

License

MIT © 2026 Hao Zhang