dingtalk-workflow-content-pipeline

Content Creation Pipeline (dingtalk-workflow-content-pipeline)

Mark content ideas in the conversation to automatically research topics, check for duplicate topics, assess value, and generate a complete proposal. Compared with the Feishu version, DingTalk group chat replaces Slack, and dws devdoc adds DingTalk developer documentation as a research source.

Trigger Method

"内容创意"  "灵感记录"  "选题评估"  "内容方案"  "创意管道"  "想写一篇关于"

Assessment Dimensions

Dimension	Weight	Scoring Criteria
Timeliness	25%	Whether the topic is currently trending (1-10)
Differentiation	25%	Whether there is a unique angle (1-10)
Audience Value	20%	How helpful it is to the target audience (1-10)
Executability	15%	Whether materials are abundant and whether it is easy to produce (1-10)
Extensibility	15%	Whether it can be split into a series of content (1-10)

Total Score = Timeliness×0.25 + Differentiation×0.25 + Audience Value×0.20 + Executability×0.15 + Extensibility×0.15

Data Initialization

On first use, automatically create the aitable data tables:

# 创建 Base
dws aitable base create --name "内容创作管道" --yes -f json
# 返回 baseId，记为 <cp-base-id>

# 创建创意库表
dws aitable table create --base-id <cp-base-id> --name "创意库" --yes -f json
# 返回 tableId，记为 <ideas-table-id>

# 创建创意研究表
dws aitable table create --base-id <cp-base-id> --name "创意研究" --yes -f json
# 返回 tableId，记为 <research-table-id>

# 添加业务字段（表创建后自动带一个"标题"primaryDoc字段，需额外添加业务字段）
dws aitable field create --base-id <cp-base-id> --table-id <ideas-table-id> --fields '[{"fieldName":"创意标题","type":"text"},{"fieldName":"来源","type":"text"},{"fieldName":"话题领域","type":"text"},{"fieldName":"评估分数","type":"number"},{"fieldName":"时效性","type":"number"},{"fieldName":"差异化","type":"number"},{"fieldName":"受众价值","type":"number"},{"fieldName":"可执行性","type":"number"},{"fieldName":"延展性","type":"number"},{"fieldName":"状态","type":"text"},{"fieldName":"关联任务ID","type":"text"},{"fieldName":"创建时间","type":"date"}]' --yes -f json

dws aitable field create --base-id <cp-base-id> --table-id <research-table-id> --fields '[{"fieldName":"关联创意","type":"text"},{"fieldName":"研究来源","type":"text"},{"fieldName":"研究类型","type":"text"},{"fieldName":"摘要","type":"text"},{"fieldName":"相关性","type":"text"}]' --yes -f json

Table 1: Idea Library (idea_library)

Field Name	Type	Description
Idea Title	Text	Content Topic
Source	Text	Conversation/Group Chat/Manual/WebSearch
Topic Domain	Text	Technology/Business/Life/Education/Other
Assessment Score	Number	0-10 (weighted across five dimensions)
Timeliness	Number	1-10
Differentiation	Number	1-10
Audience Value	Number	1-10
Executability	Number	1-10
Extensibility	Number	1-10
Status	Text	Inspiration/Researching/Proposal/In Progress/Completed
Associated Task ID	Text	DingTalk To-Do taskId
Creation Time	Date	Time of entry

Table 2: Idea Research (idea_research)

Field Name	Type	Description
Associated Idea	Text	Associated idea title
Research Source	Text	URL or description
Research Type	Text	Topic Trends/Competitor Analysis/User Discussion/Technical Documentation/Other
Summary	Text	Research findings
Relevance	Text	High/Medium/Low

Working Modes

Mode 1: Idea Assessment (Complete Process)

Trigger: "I have a content idea, and I want to write an article about AI Agent implementation in enterprises" / "I want to write an article about XX"

Workflow:

Step 1: 解析创意
├── 提取话题关键词
├── 判断话题领域
└── 确定目标平台（文章/视频/推文等）

Step 2: 并行研究
├── WebSearch → 搜索话题最新动态和讨论热度
├── dws devdoc article search --keyword "<话题>" → 搜索钉钉开发文档（技术类创意适用）
└── 检查知识库重复选题（如 knowledge-base 已初始化）
    └── dws aitable record query --base-id <kb-base-id> --keyword "<关键词>"

Step 3: 评估创意价值
├── 五个维度分别打分
├── 计算加权总分
└── 生成评估报告

Step 4: 生成内容方案
├── 标题建议（3-5 个）
├── 目标字数/时长
├── 详细大纲
├── 开头建议
└── 目标受众

Step 5: 保存创意并创建待办
├── dws aitable record create → 保存创意库
├── dws aitable record create → 保存研究成果
└── dws todo task create → 创建跟踪待办

Output format:

## 创意评估

### 话题：AI Agent 在企业落地

### 评估
| 维度 | 评分 | 说明 |
|------|------|------|
| 时效性 | 8/10 | AI Agent 是当下热门话题 |
| 差异化 | 6/10 | 已有不少类似内容，需找到独特角度 |
| 受众价值 | 9/10 | 企业用户高度关注 |
| 可执行性 | 8/10 | 案例丰富，易于产出 |
| 延展性 | 7/10 | 可拆分为系列文章 |
| **总分** | **7.6/10** | |

### 内容方案
- 标题建议：① AI Agent 落地实战：5 个真实案例拆解
- 类型：深度文章
- 目标字数：3000-5000 字
- 大纲：
  1. AI Agent 不是聊天机器人
  2. 案例一：自动化客服
  3. 案例二：代码审查
  ...

### 重复检查
知识库中未发现类似选题。

✓ 创意已保存，待办已创建。

Mode 2: Quick inspiration capture

Trigger: "Save an inspiration: a topic about XX" / "Save this idea"

Workflow:

Step 1: 记录创意标题和简要描述
└── dws aitable record create → 保存到创意库（状态=灵感）

Step 2: 输出确认

Mode 3: Idea library browsing

Trigger: "Show me my idea library" / "What topics are there" / "Idea list"

Workflow:

Step 1: 查询创意库
└── dws aitable record query --base-id <cp-base-id> --table-id <ideas-table-id> --limit 50 --yes -f json

Step 2: 按评估分数排序，格式化输出

Output format:

## 创意库清单

| 标题 | 领域 | 评分 | 状态 | 创建时间 |
|------|------|------|------|---------|
| AI Agent 落地实战 | 科技 | 7.6 | 执行中 | 03-28 |
| 飞书自动化指南 | 科技 | 8.2 | 方案 | 03-25 |
| 投资入门系列 | 商业 | 6.5 | 灵感 | 03-20 |

共 N 个创意，其中 M 个待执行。

Mode 4: Idea upgrade (inspiration → research → proposal)

Trigger: "Upgrade the idea for XXX" / "Research this topic XXX"

Workflow:

Step 1: 查询指定创意
└── dws aitable record query --keyword "<创意标题>"

Step 2: 执行研究流程（同模式 1 的 Step 2）
├── WebSearch → 话题动态
├── devdoc → 技术文档
└── 知识库 → 重复检查

Step 3: 更新创意评估分数和状态
└── dws aitable record update --records '[{"recordId":"...","cells":{"评估分数":7.6,"状态":"研究中"}}]' --yes

Step 4: 如用户确认，生成完整方案并更新状态为"方案"

Mode 5: Idea execution tracking

Trigger: "Is XXX finished" / "Idea execution status"

Workflow:

Step 1: 查询创意库中"执行中"状态的创意
└── dws aitable record query --keyword "执行中"

Step 2: 检查关联的待办任务状态
└── dws todo task list

Step 3: 生成执行进度报告

dws CLI command reference

Create idea pipeline

dws aitable base create --name "内容创作管道" --yes -f json
dws aitable table create --base-id <cp-base-id> --name "创意库" --yes -f json
dws aitable table create --base-id <cp-base-id> --name "创意研究" --yes -f json

Save idea

dws aitable record create --base-id <cp-base-id> --table-id <ideas-table-id> --records '[{"cells":{"标题":"AI Agent落地实战","创意标题":"AI Agent落地实战","来源":"对话","话题领域":"科技","评估分数":7.6,"状态":"灵感","创建时间":"2026-04-06"}}]' --yes -f json

Save research results

dws aitable record create --base-id <cp-base-id> --table-id <research-table-id> --records '[{"cells":{"标题":"AI Agent落地实战-话题动态研究","关联创意":"AI Agent落地实战","研究来源":"https://...","研究类型":"话题动态","摘要":"...","相关性":"高"}}]' --yes -f json

Query idea library

dws aitable record query --base-id <cp-base-id> --table-id <ideas-table-id> --limit 50 --yes -f json

Update idea status

dws aitable record update --base-id <cp-base-id> --table-id <ideas-table-id> --records '[{"recordId":"rec_xxx","cells":{"状态":"执行中","关联任务ID":"todo_xxx"}}]' --yes -f json

Create tracking to-do

dws todo task create --title "创作「AI Agent落地实战」" --priority "20" --due "2026-04-15T18:00:00+08:00" --yes -f json

Error handling: dws todo task create The server may occasionally return an NPE error (errorCode 999); if it fails, it is recommended to retry once.

Search DingTalk Open Platform documentation (research reference)

dws devdoc article search --keyword "AI Agent" --yes -f json

Send group message notification

dws chat message send-by-bot --robot-code "bot_code" --group "openConversationId" --title "创意管道更新" --text "新创意：AI Agent落地实战\n评分：7.6/10" --yes

Linked Skill

Linked Skill	Data interaction	Description
knowledge-base	Search to check for duplicate topics	Avoid duplicate content
business-advisor	Provide idea data	Advisory group analyzes content strategy
social-tracker	Provide social performance data	Evaluate the potential contribution of ideas to social growth
multi-agent-dev	Execute idea content creation	Multi-agent parallel writing

Notes

All dws commands must include the --yes flag
Idea evaluation scores retain one decimal place
Status flow: Inspiration → Researching → Proposal → In progress → Completed
dws devdoc Only searches DingTalk Open Platform documentation, not a general search engine
During the research phase, use WebSearch to obtain broader topic trends
dws todo task create The server occasionally returns NullPointerException; retry once on failure, and if it still fails, only save it in the idea library for later manual creation
dws devdoc article search Only searches DingTalk Open Platform documentation; returned results include title/url/desc and can be used for technical idea research

Known issues

dws todo task create Occasional server NPE (errorCode 999), requires retry

About this skill

Trigger Method

Assessment Dimensions

Data Initialization

Working Modes

Mode 1: Idea Assessment (Complete Process)

Mode 2: Quick inspiration capture

Mode 3: Idea library browsing

Mode 4: Idea upgrade (inspiration → research → proposal)

Mode 5: Idea execution tracking

dws CLI command reference

Create idea pipeline

Save idea

Save research results

Query idea library

Update idea status

Create tracking to-do

Search DingTalk Open Platform documentation (research reference)

Send group message notification

Linked Skill

Notes

Known issues