Token导航 LogoToken导航

Agent Skill · 研究检索

todokan(todokan 搜索)

todokan 用于查找、检索和筛选相关信息,适合在 OpenClaw 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

todokan 是研究检索类 Agent Skill,可用宿主为 OpenClaw,通过ClawHub获取,权限为调用外部服务;许可证为 MIT-0,第三方扫描已通过,按安装量在研究检索类的 73,675 个 Skill 里排第 9,392。

总安装 14,302周安装 590GitHub Stars 公开资料未说明下载 4,673核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
MIT-0
版本
1.3.2

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:todokan(todokan 搜索)
来源仓库:https://github.com/banatealab/todokan
安装命令:
openclaw skills install todokan
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 OpenClaw 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

ClawHubOpenClaw
openclaw skills install todokan

这个 Skill 是做什么的

研究检索类 Skill 主要解决「找资料、核对事实、把来源整理清楚」的问题,通常会让助手先检索再总结。

通过 MCP 在 Todokan 中管理任务、看板、想法和评论。适用于敏捷开发和团队协作中的任务看板管理场景。安装后可通过 openclaw skills install todokan 使用,需确认宿主环境是否支持 MCP 协议。使用前请检查权限范围,避免触发不必要的网络请求或数据修改操作。

建议结合具体用例查阅原始仓库文档,确保与 Todokan 平台兼容且功能完整。

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 OpenClaw;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 openclaw skills install todokan;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它需要调用外部服务,先在测试目录或测试账号里跑,确认没问题再放到正式项目。

常见问题

6 条

todokan 怎么安装?

在终端执行 openclaw skills install todokan。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

todokan 支持哪些 AI 工具?

公开资料标注支持 OpenClaw。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

todokan 需要什么权限?

权限级别是调用外部服务。该 Skill 可能调用第三方服务、云服务或外部模型 API,使用前需要确认账号、额度、数据发送范围和服务条款。

todokan 是免费的吗?

来源标注的许可证是 MIT-0。本站不代收任何费用,也不托管安装包;能否商用、要不要署名,以许可证原文为准。

todokan 用的人多吗?

公开数据是累计安装 14,302 次,近一周安装 590 次,下载 4,673 次。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

todokan 安全吗?

第三方扫描结果为通过,但这只覆盖已知的静态检查项,不能替代自己读一遍源码和安装命令。

值得选的地方和要先想清楚的地方

值得选的地方

  • 专注 OpenClaw 一种宿主,触发方式和使用习惯不用来回切换。
  • 许可证明确为 MIT-0,商用和分发边界好判断。
  • 第三方扫描已通过,常见的高风险行为没有命中。

先想清楚的地方

  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 OpenClaw,想直接扩展它能力的人。
  • 需要处理研究检索类任务、又不想每次都手写提示词的人。
  • 需要许可证清楚、并且有第三方扫描结果的团队项目。

不太适合

  • 需要厂商级技术支持和 SLA 的生产系统。

研究检索类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
find-skills

寻找技能 · Codex、Claude

31,199,81316,486操作浏览器MIT有提醒
self-improving-agent

自我改进 Agent · OpenClaw

16,464,0023,396接触敏感数据MIT-0通过
soultrace

灵魂追踪 · Codex、Claude

7,588,94412需要联网未明确有提醒
self-improving

自我完善 · OpenClaw

6,829,2841,046需要联网MIT-0通过
microsoft-foundry

微软代工厂 · Codex、Claude

6,614,144764调用外部服务未明确有提醒
todokan

todokan 搜索 · 本页

14,302—调用外部服务MIT-0通过

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

name
todokan
version
1.3.2
description
Manage tasks, boards, thoughts, and reviews in Todokan via MCP
homepage
https://todokan.com
metadata
{"category":"productivity","tags":["tasks","kanban","mcp","planning","review"],"requires":{"env":["TODOKAN_API_KEY","TODOKAN_MCP_URL"],"mcp":true},"openclaw":{"homepage":"https://todokan.com","requires":{"env":["TODOKAN_API_KEY","TODOKAN_MCP_URL"]}}}

Todokan — Kanban Task Management

Todokan is a kanban-style task manager. You can manage the user's tasks, boards, and projects through MCP tools.

Prerequisites

  • A Todokan MCP server must be available (see README for setup)
  • Required env vars: TODOKAN_API_KEY, TODOKAN_MCP_URL (declared in skill metadata)

Trigger — When to Activate This Skill

Activate the Todokan skill when the user has one of the following intents:

IntentExample
Create / edit / delete a task"Create a task: review PR"
Show boards or tasks"Show me my tasks", "What's on the dev board?"
Change status"Mark task X as done"
Save research results"Save this as a task / document in Todokan"
Briefing / summary from tasks"Give me a briefing of my open tasks"
Attach a document to a task"Write a note on task X"
Search topics across boards"What did I note about the investor meeting?"
Retrieve changes since last check"What's new since this morning?"

Do not activate when the user is just talking about tasks in general without referencing Todokan.


Tool Ordering

Follow this order to achieve consistent results:

Reading (always orient first)

1. list_habitats            → Which workspaces exist?
2. list_boards              → Which boards exist? (note the IDs)
3. list_tasks               → Tasks on a board (with filters)
4. search_across_habitats   → Full-text search across boards/habitats
5. get_events_since         → Retrieve changes since a timestamp
6. list_board_labels        → Available labels + usage counts
7. list_task_documents      → Documents attached to a task
8. read_document            → Content of a document

Writing (only after orientation + confirmation)

9.  create_task / create_board / create_habitat
10. update_task / update_task_by_title
11. create_document / add_document_to_task
12. delete_task              → Only after explicit confirmation

AI-Assisted (optional)

13. propose_task_variants    → Generate 2-3 variants
14. confirm_task_fields      → Review fields before creation

Golden rule: Never write blindly. Always call list_boards first to discover IDs — never guess UUIDs.


Mandatory Checks

Before executing write actions, clarify the following:

Before create_task

  • [ ] Board: Which board? (If unclear → show list_boards, let the user choose)
  • [ ] Title: Short, precise, imperative (max 80 characters)
  • [ ] Priority: low / normal / high — default to normal if unclear
  • [ ] Due date: Only set if mentioned by the user

Before update_task

  • [ ] Correct task? State the title + board for confirmation
  • [ ] Which fields? Only change the fields the user requested

Before delete_task

  • [ ] Always ask for explicit confirmation — "Should I permanently delete task '[Title]' on board '[Board]'? This cannot be undone."

Before create_document

  • [ ] Clarify format: markdown, text, or html
  • [ ] Link: Attach to which task (or standalone)?

Guardrails

No Hallucination

  • Use only real data from MCP tool responses. Never fabricate task IDs, board names, or content.
  • If a tool call fails or returns empty data: inform the user, do not improvise.
  • When in doubt, show the actual results and ask.

No Sensitive Data

  • Do not store passwords, API keys, tokens, or personal data in task titles or descriptions.
  • If the user mentions sensitive information, warn them: "This may contain sensitive data — should I really store it in Todokan?"

Source Attribution

  • When storing content from external research (web, files, other tools) in Todokan, note the source in the task description or document:
  Source: [URL or filename]
  Created by: Agent on [date]

Scope Awareness

  • Worker endpoint (/mcp-worker): Read-only + comments (add_comment). No task/board CUD, no document creation.
  • Planner endpoint (/mcp): Full access. Still ask before destructive actions.
  • On scope errors: explain to the user that the current endpoint does not have the required permissions.

Idempotency

  • On network errors: do not blindly retry the same action. First check whether the action was already performed (list_tasks).

Output Format

Briefing (summary of existing tasks)

## Briefing: [Board Name] — [Date]

**Open (todo):** X tasks
**In progress (doing):** Y tasks
**Completed (done):** Z tasks

### Urgent (high priority)
- [ ] [Task Title] — due [Date]
- [ ] [Task Title] — due [Date]

### In Progress
- [~] [Task Title] — since [Date]

### Next Steps
[1-2 sentences recommendation based on the data]

Draft (preview before task creation)

## Task Draft

| Field       | Value                         |
|-------------|-------------------------------|
| Board       | [Board Name]                  |
| Title       | [Title, max 80 chars]         |
| Description | [Description, max 500 chars]  |
| Status      | todo                          |
| Priority    | [low / normal / high]         |
| Due         | [YYYY-MM-DD or —]             |
| Labels      | [label1, label2]              |

Should I create this task?

Document Draft

## Document Draft

**Title:** [Title]
**Format:** markdown
**Linked to:** [Task Title] on [Board Name]

---
[Document content]
---

Should I create this document?

Data Model

Habitat (workspace/project)
  └── Board (kanban board, type: "task" or "thought")
       └── Task (individual item with status, priority, labels, due date)
            └── Document (attached notes/docs in markdown, text, or html)
  • Habitats group boards. A user can have multiple habitats.
  • Boards are kanban boards. Type task for actionable items, thought for ideas/notes.
  • Tasks live on a board and move through status columns.
  • Documents are rich text attached to tasks.

Status Values

StatusMeaning
todoNot started
doingIn progress
doneCompleted

Priority Values

PriorityMeaning
lowLow priority
normalDefault priority
highHigh/urgent priority

Available MCP Tools

Reading Data

  • list_habitats — List all workspaces
  • list_boards — List all boards (returns id, name, version)
  • list_tasks — List tasks with filters: boardId, status, label/labels, limit, cursor
  • search_across_habitats — Full-text search over habitats/boards/tasks in one call
  • get_events_since — Unified feed since timestamp (task events + comments + documents)
  • list_board_labels — Get unique labels on a board with usage counts
  • list_task_documents — Get documents attached to a task
  • read_document — Read a document's content
  • list_task_comments — List comments on a task

Creating & Modifying (Planner endpoint only)

  • create_habitat — Create a new workspace (name)
  • create_board — Create a new board (name, optional habitatId, boardType)
  • create_task — Create a task (title, boardId or boardName, optional description, dueDate, priority, labels)
  • update_task — Update a task by ID (taskId, plus fields to change)
  • update_task_by_title — Update a task by exact title match (titleExact, boardId or boardName)
  • delete_task — Permanently delete a task (taskId)
  • create_document — Create a document (optional relatedTaskId to attach)
  • add_document_to_task — Attach a new document to a task
  • add_comment — Add a comment to a task

AI-Assisted Creation

  • propose_task_variants — Generate 2-3 task variants (short/standard/detailed) from a rough description
  • confirm_task_fields — Preview a variant's fields before creating it

AI Visibility Gate

By default, the MCP server only returns tasks where aiEnabled: true. Tasks with aiEnabled: false are invisible to MCP agents — they will not appear in list_tasks, search_across_habitats, get_events_since, or get_task.

Users control this via a "Send to AI" button on each task card. When clicked, it sets aiEnabled: true, assignee: 'ai', and status: 'doing'.

  • To see only AI-assigned tasks: list_tasks { "assignee": "ai" }
  • To see all AI-enabled tasks: list_tasks {} (default — only AI-enabled tasks are returned)
  • To explicitly include non-AI tasks: list_tasks { "aiEnabled": false } (override, useful for reporting)

OAuth & Authentication

  • OAuth 2.1 with PKCE (RS256 JWT)
  • Token lifetime: 30 days, no refresh token
  • Planner (/mcp): Full CRUD access — all scopes
  • Worker (/mcp-worker): boards:read, tasks:read, labels:read, docs:read, comments:read, comments:write
  • No rate limiting — still be conservative with calls
  • Activity logging: Every tool call is logged server-side

Common Workflows

List all tasks on a board

  1. list_boards to find the board ID
  2. list_tasks with boardId to get tasks

Create a task

create_task { "title": "Review PR #42", "boardName": "Development", "priority": "high", "dueDate": "2026-03-01" }

Move a task to done

update_task { "taskId": "<uuid>", "status": "done" }

Add labels to a task

update_task { "taskId": "<uuid>", "labels": ["bug", "frontend"] }

Find tasks by label

list_tasks { "boardId": "<uuid>", "labels": ["bug"] }

Search across all habitats

search_across_habitats { "query": "investor meeting", "limit": 20 }

Poll event feed (agent loop)

get_events_since { "since": "2026-02-24T08:00:00Z", "limit": 200 }

Tips

  • Always call list_boards first to discover available board IDs — don't guess UUIDs.
  • Use boardName instead of boardId when the user refers to boards by name.
  • Due dates use YYYY-MM-DD format.
  • Task titles are max 80 characters, descriptions max 500 characters.
  • Labels are free-form strings (max 10 per task).
  • The update_task tool requires the task's UUID. Use list_tasks to find it, or update_task_by_title if you only know the title.

平台分布

OpenClaw

97.29%
按下载量换算4,546

安全审计

VirusTotal

通过

ClawScan

通过

Static analysis

未展示

权限和风险

调用外部服务

该 Skill 可能调用第三方服务、云服务或外部模型 API,使用前需要确认账号、额度、数据发送范围和服务条款。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。目前只有一个公开来源,建议同时核对源仓库。

继续浏览同类 Skills