← 返回 Skills 市场
pdurlej

Things Cloud

作者 Piotr Durlej · GitHub ↗ · v0.1.0 · MIT-0
cross-platform ✓ 安全检测通过
38
总下载
0
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install things-cloud
功能描述
Manage Things 3 tasks through Things Cloud using the maintained things-cloud-sdk CLI and MCP server, with dry-run safety for agent writes.
使用说明 (SKILL.md)

Things Cloud

Use this skill when the user wants an agent to inspect or safely update Things 3 tasks through Things Cloud.

Prefer the MCP server when the host supports MCP. Use the CLI as the fallback or when the user asks for explicit shell commands.

Setup

Install the maintained CLI and MCP server:

go install github.com/pdurlej/things-cloud-sdk/cmd/[email protected]
go install github.com/pdurlej/things-cloud-sdk/cmd/[email protected]

Set credentials:

export THINGS_USERNAME="[email protected]"
export THINGS_TOKEN="your-things-cloud-password-or-token-alias"

THINGS_PASSWORD also works when THINGS_TOKEN is not set. Do not store credentials in the repository or in skill files.

MCP Configuration

Use this stdio MCP server config for OpenClaw, Claude Code, Codex hosts, or any MCP-compatible agent runtime:

{
  "mcpServers": {
    "things": {
      "command": "things-mcp",
      "env": {
        "THINGS_USERNAME": "[email protected]",
        "THINGS_TOKEN": "your-things-cloud-password-or-token-alias"
      }
    }
  }
}

Agent Policy

  • Read before writing.
  • Prefer MCP tools for task operations.
  • Use CLI --simple output for compact task lists.
  • Use --dry-run before any user-visible write generated by an agent.
  • Summarize the planned change and get confirmation before non-dry-run writes.
  • Never write to the local Things SQLite database.
  • Do not infer completion from absence in active views; use completed/logbook.
  • Use task UUIDs returned by Things Cloud output. Do not invent UUIDs.

Common Reads

With MCP:

  • list_tasks with view set to today, inbox, anytime, someday, or upcoming
  • search_tasks for title/note search
  • list_projects, list_areas, and list_tags for metadata

With CLI:

things-cloud-cli today --simple
things-cloud-cli inbox --simple
things-cloud-cli anytime --simple
things-cloud-cli someday --simple
things-cloud-cli upcoming --simple
things-cloud-cli search "invoice" --simple

Completion Evidence

Use completed/logbook commands when the user asks whether something was done:

things-cloud-cli completed --since 2026-05-20T00:00:00Z --format full
things-cloud-cli logbook --since 2026-05-20 --limit 50 --format full

Safe Writes

Preview first:

things-cloud-cli create "Follow up with Marta" --when today --dry-run
things-cloud-cli edit \x3Ctask-uuid> --title "New title" --dry-run
things-cloud-cli complete \x3Ctask-uuid> --dry-run

After user confirmation, run the same command without --dry-run.

Recurring tasks:

things-cloud-cli create "Check car listings" --repeat every-day --dry-run
things-cloud-cli create "Weekly review" --repeat weekly:mon --dry-run
things-cloud-cli create "Follow up after reply" --repeat after-completion:every-day --dry-run

Clear recurrence:

things-cloud-cli edit \x3Ctask-uuid> --repeat none --dry-run

Output Rules

  • Keep raw command output as JSON.
  • Summaries should mention the task title, UUID, view/status, and whether the write was dry-run or executed.
  • If a task reference is ambiguous, search first and ask the user to choose.
  • If credentials are missing, ask the user to set THINGS_USERNAME and either THINGS_TOKEN or THINGS_PASSWORD.
安全使用建议
Install only if you are comfortable giving an agent access to your Things Cloud account and task data. Prefer THINGS_TOKEN over THINGS_PASSWORD when possible, keep credentials out of repositories, review dry-run output carefully, and require explicit confirmation before any real task creation, edit, completion, or deletion.
能力评估
Purpose & Capability
The task read/write capability matches the stated Things Cloud task-management purpose, and the artifact explicitly distinguishes reads, dry-run previews, and confirmed writes.
Instruction Scope
Runtime instructions require reading before writing, using dry-run for agent-generated writes, confirming before non-dry-run changes, using returned UUIDs, and asking when task references are ambiguous.
Install Mechanism
Installation is disclosed as pinned Go installs for things-cloud-cli and things-mcp, with no hidden scripts or executable artifact files; pkg.go.dev also documents the same CLI/MCP package and pinned v0.2.3 install path. ([pkg.go.dev](https://pkg.go.dev/github.com/pdurlej/things-cloud-sdk%40v0.2.3))
Credentials
The skill requires Things Cloud account credentials and can read or modify personal task data, but that access is expected for the integration and is clearly disclosed through THINGS_USERNAME, THINGS_TOKEN, and THINGS_PASSWORD.
Persistence & Privilege
No background persistence is installed by the skill itself; it discloses optional config and CLI read-state cache paths, while the underlying package documents credential config and cache behavior. ([pkg.go.dev](https://pkg.go.dev/github.com/pdurlej/things-cloud-sdk%40v0.2.3))
如何使用
  1. 确保已安装 OpenClaw(本地或 Docker 部署)
  2. 在对话框中输入安装命令:/install things-cloud
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /things-cloud 触发
  4. 根据 Skill 的参数说明提供必要输入,即可获得结构化输出
版本历史
v0.1.0
Initial Things Cloud skill for safe OpenClaw, Codex, and Claude Code task automation.
元数据
Slug things-cloud
版本 0.1.0
许可证 MIT-0
累计安装 0
当前安装数 0
历史版本数 1
常见问题

Things Cloud 是什么?

Manage Things 3 tasks through Things Cloud using the maintained things-cloud-sdk CLI and MCP server, with dry-run safety for agent writes. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 38 次。

如何安装 Things Cloud?

在 OpenClaw 或 Claude Code 对话框中运行命令「/install things-cloud」即可一键安装,无需额外配置。

Things Cloud 是免费的吗?

是的,Things Cloud 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。

Things Cloud 支持哪些平台?

Things Cloud 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(cross-platform)。

谁开发了 Things Cloud?

由 Piotr Durlej(@pdurlej)开发并维护,当前版本 v0.1.0。

💬 留言讨论