这是一个为 Clawdbot 打造的 TaskBoardAI 集成技能。它允许 Clawdbot 通过 MCP (Model Context Protocol) 协议,直接在后台管理任务看板,实现“意图识别 -> 任务创建 -> 执行 -> 归档”的全自动化工作流。
- 全生命周期管理:支持创建 (Todo)、执行 (Doing)、阻塞 (Blocked)、完成 (Done) 四种状态的自动流转。
- MCP 深度集成:使用 stdio 模式与 TaskBoard 后端直接通信,无需维护额外的 HTTP 服务。
- 内容自适应:自动修复了 TaskBoard 前端渲染
content字段而非description字段的问题。 - 条件反射协议:配合
AGENTS.md,实现通过关键字“【任务】”触发自动建卡。
本技能依赖于 TaskBoardAI。请先全局安装:
npm install -g taskboardai将本目录克隆至你的 Clawdbot 工作区:
cd ~/clawd/skills
git clone https://github.com/hyddd/taskboard_skill.git taskboard检查 SKILL.md 中的 <arg> 路径是否指向你系统中真实的 kanbanMcpServer.js 位置:
<arg>/opt/homebrew/lib/node_modules/taskboardai/server/mcp/kanbanMcpServer.js</arg>在与 Clawdbot 的对话中:
- 创建任务:“【任务】调研量化框架”
- 查询任务:“现在有哪些进行中的任务?”
- 更新状态:“那个调研任务做完了,结论是...”
为了获得最佳体验,建议在你的 AGENTS.md 中添加以下协议:
## Task Protocol (TaskBoard Integration)
**Trigger:** When user input contains "【任务】" (or you determine a task needs tracking):
### 1. Initiate (Create & Start)
1. **Call `taskboard` skill** immediately.
2. **Create Card:** In the "Todo" column.
- Title: The core task name.
- Content: Full context + "Owner: Clawdbot".
3. **Start Execution:** Immediately move the card to the "Doing" (In Progress) column.
4. **Announce:** "Task [Title] created and started."
### 2. Execute & Update
1. **Perform the work:** Use available tools (web_search, etc.) to fulfill the task.
2. **If Blocked:**
- Move card to "Blocked" column.
- Update Content: Add a section `**⚠️ BLOCKER:** [Reason]`.
- Notify user of the blocker.
3. **If Successful:**
- Update Content: Append the **FINAL RESULT/SUMMARY** to the card content. This is crucial for future reference.
- Move card to "Done" column.
- Report completion to user with a brief summary.This is a TaskBoardAI integration skill designed for Clawdbot. It enables Clawdbot to manage a Kanban board directly via the MCP (Model Context Protocol), achieving a fully automated workflow of "Intent Recognition -> Task Creation -> Execution -> Archiving".
- Full Lifecycle Management: Supports automatic transitions between Todo, Doing, Blocked, and Done states.
- Deep MCP Integration: Communicates directly with the TaskBoard backend using stdio mode, eliminating the need for extra HTTP services.
- Content Adaptation: Automatically fixes the issue where the TaskBoard frontend renders the
contentfield instead ofdescription. - Reflex Protocol: Works with
AGENTS.mdto trigger automatic card creation via the keyword "【任务】" (Task).
This skill depends on TaskBoardAI. Please install it globally first:
npm install -g taskboardaiClone this directory into your Clawdbot workspace:
cd ~/clawd/skills
git clone https://github.com/hyddd/taskboard_skill.git taskboardCheck SKILL.md to ensure the <arg> path points to the actual kanbanMcpServer.js location on your system:
<arg>/opt/homebrew/lib/node_modules/taskboardai/server/mcp/kanbanMcpServer.js</arg>In your conversation with Clawdbot:
- Create Task: "【任务】Research quantitative frameworks"
- Query Task: "What tasks are currently in progress?"
- Update Status: "The research task is done, the conclusion is..."
For the best experience, it is recommended to add the following protocol to your AGENTS.md:
## Task Protocol (TaskBoard Integration)
**Trigger:** When user input contains "【任务】" (or you determine a task needs tracking):
### 1. Initiate (Create & Start)
1. **Call `taskboard` skill** immediately.
2. **Create Card:** In the "Todo" column.
- Title: The core task name.
- Content: Full context + "Owner: Clawdbot".
3. **Start Execution:** Immediately move the card to the "Doing" (In Progress) column.
4. **Announce:** "Task [Title] created and started."
### 2. Execute & Update
1. **Perform the work:** Use available tools (web_search, etc.) to fulfill the task.
2. **If Blocked:**
- Move card to "Blocked" column.
- Update Content: Add a section `**⚠️ BLOCKER:** [Reason]`.
- Notify user of the blocker.
3. **If Successful:**
- Update Content: Append the **FINAL RESULT/SUMMARY** to the card content. This is crucial for future reference.
- Move card to "Done" column.
- Report completion to user with a brief summary.