lark-workflow-crm

Micro Team CRM Workflow

CRITICAL — Before starting, MUST first use the Read tool to read ../lark-shared/SKILL.md, which contains authentication, permission handling

Applicable Scenarios

"Help me create a CRM system" / "Set up sales management" / "Create new CRM"
"Add customer: Beijing Starlight Technology, contact Mr. Wang" / "Customer record creation"
"Create opportunity for Beijing Starlight: ERP procurement, 500k" / "Opportunity follow-up"
"Follow-up record: called Mr. Wang today, scheduled an in-person visit next Wednesday"
"Show me the sales funnel" / "How much was won this month" / "Performance ranking"
"Generate today's sales daily report" / "This week's follow-up summary"
"Create customer registration form" / "Inquiry form"
"Create sales kanban" / "Sales dashboard"
"Follow-up reminder" / "Overdue alert" / "Deal won celebration" / "Weekly sales briefing"

Prerequisites

Only supports user identity. Before execution, ensure authorization has been granted:

lark-cli auth login --domain base,im,contact

System Architecture

微型团队 CRM (lark-workflow-crm)
│
├── 数据层：飞书多维表格 (Bitable)
│   ├── 客户信息主表 (customers)        — 客户档案
│   ├── 商机管理表 (opportunities)      — 销售漏斗
│   └── 跟进记录表 (follow_ups)         — 沟通记录
│
├── 计算层
│   ├── 赢单概率建议 (formula)          — 根据商机阶段自动建议概率
│   ├── 商机天数 (formula)              — 距商机创建的天数
│   ├── 客户商机数量 (formula)          — 关联商机计数
│   ├── 客户累计赢单金额 (formula)       — 关联赢单金额汇总
│   └── 商机跟进次数 (formula)          — 关联跟进计数
│
├── 关联关系（双向）
│   ├── customers ↔ opportunities       — 一个客户多个商机
│   ├── customers ↔ follow_ups          — 跟进记录关联客户
│   └── opportunities ↔ follow_ups      — 跟进记录关联商机
│
├── 视图层
│   ├── 商机阶段看板 (Kanban)           — 按阶段分列拖拽
│   ├── 我的商机                        — 当前用户负责的商机
│   ├── 跟进时间线                      — 按时间倒序的跟进记录
│   └── 本月赢单                        — 当月赢单列表
│
├── 仪表盘
│   ├── 销售漏斗 (funnel)               — 各阶段商机数量
│   ├── 本月赢单金额 (statistics)        — SUM WHERE 阶段=赢单
│   ├── 商机总金额 (statistics)          — 在途商机金额
│   ├── 客户数量 (statistics)            — 总客户数
│   ├── 阶段分布 (pie)                  — 各阶段占比
│   ├── 负责人业绩排行 (column)          — 赢单金额排行
│   └── 本周跟进 (table)                — 本周跟进记录
│
├── 表单
│   └── 客户咨询登记表                  — 客户自助提交咨询
│
├── 交互层：用户自然语言指令
│   ├── 系统初始化 → 创建 Base + 表 + 字段 + 关联 + 视图 + 仪表盘 + 表单
│   ├── 日常操作 → 客户建档 / 商机管理 / 跟进记录
│   └── 数据分析 → 漏斗查询 / 业绩统计 / 销售日报
│
└── 通知层：飞书消息
    ├── 每日跟进提醒
    ├── 超时未跟进预警
    ├── 赢单报喜通知
    └── 每周销售简报

Agent Core Rules

Only use lark-cli base +... atomic commands — do not go through lark-cli api for direct calls
When creating a table, you must first create an empty table and then add fields one by one — +table-create without --fields parameter, first create an empty table and then use +field-create to add fields one by one, with an interval of 1.5 seconds each time
Link fields must use bidirectional links — bidirectional: true, otherwise cross-table aggregation cannot be done in formulas
For cross-table calculations, prefer using formula fields — formula fields are a superset of lookup, so use formula instead of lookup whenever possible
Before creating formula fields, first read the guide — read the ../lark-base/references/formula-field-guide.md
+data-query table-id and put it in the DSL — does not support --table-id parameter, it needs to be placed in --dsl of datasource.table.tableId in
+data-query only supports stored fields — does not support formula fields, lookup fields, or link fields for aggregation
Before writing records, first get the field structure — +field-list to confirm field names and types
Table names/field names must match exactly — guessing based on natural language is prohibited, they must come from +table-list / +field-list return values
+xxx-list Concurrency is prohibited — all list commands are executed serially
Writes serial + interval — +record-upsert serially, with an interval of 0.5-1 second
Paginated retrieval — +record-list --limit 200, using offset pagination
When stage changes are mentioned in follow-up records, update the opportunity synchronously — when the user says "advance to proposal quotation", update the corresponding opportunity's opportunity stage field
Before creating a dashboard, first read the guide — read the ../lark-base/references/dashboard-block-data-config.md
Before creating a form, first read the guide — read the ../lark-base/references/lark-base-form-create.md

Phase 1: System Initialization

Step 1.1: Create Base

lark-cli base +base-create --name "微型团队CRM" --as user

Record the returned base_token(needed by all subsequent commands).

Step 1.2: Create data tables (create empty tables first, with a 2-second interval)

IMPORTANT: do not use the --fields parameter to create multiple fields at once; this easily triggers rate limiting.

# 表 1：客户信息主表
lark-cli base +table-create --base-token "<base_token>" --name "客户信息" --as user
sleep 2

# 表 2：商机管理表
lark-cli base +table-create --base-token "<base_token>" --name "商机管理" --as user
sleep 2

# 表 3：跟进记录表
lark-cli base +table-create --base-token "<base_token>" --name "跟进记录" --as user

Step 1.3: Get the table_id of each table

lark-cli base +table-list --base-token "<base_token>" --as user

Record each table's table_id。

Step 1.4: Add storage fields one by one (with a 1.5-second interval between each creation)

Customer Info table fields

lark-cli base +field-create --base-token "<base_token>" --table-id "<customers_table_id>" --json '{"type":"text","name":"客户名称"}' --as user
sleep 1.5
lark-cli base +field-create --base-token "<base_token>" --table-id "<customers_table_id>" --json '{"type":"select","name":"客户类型","multiple":false,"options":[{"name":"个人客户"},{"name":"企业客户"}]}' --as user
sleep 1.5
lark-cli base +field-create --base-token "<base_token>" --table-id "<customers_table_id>" --json '{"type":"text","name":"联系人"}' --as user
sleep 1.5
lark-cli base +field-create --base-token "<base_token>" --table-id "<customers_table_id>" --json '{"type":"text","name":"手机号","style":{"type":"phone"}}' --as user
sleep 1.5
lark-cli base +field-create --base-token "<base_token>" --table-id "<customers_table_id>" --json '{"type":"text","name":"邮箱","style":{"type":"email"}}' --as user
sleep 1.5
lark-cli base +field-create --base-token "<base_token>" --table-id "<customers_table_id>" --json '{"type":"select","name":"行业","multiple":false,"options":[{"name":"互联网"},{"name":"金融"},{"name":"教育"},{"name":"制造业"},{"name":"零售"},{"name":"其他"}]}' --as user
sleep 1.5
lark-cli base +field-create --base-token "<base_token>" --table-id "<customers_table_id>" --json '{"type":"select","name":"公司规模","multiple":false,"options":[{"name":"1-10人"},{"name":"11-50人"},{"name":"51-200人"},{"name":"200人以上"}]}' --as user
sleep 1.5
lark-cli base +field-create --base-token "<base_token>" --table-id "<customers_table_id>" --json '{"type":"select","name":"来源渠道","multiple":false,"options":[{"name":"主动咨询"},{"name":"转介绍"},{"name":"展会"},{"name":"线上广告"},{"name":"社群"},{"name":"其他"}]}' --as user
sleep 1.5
lark-cli base +field-create --base-token "<base_token>" --table-id "<customers_table_id>" --json '{"type":"text","name":"地址"}' --as user
sleep 1.5
lark-cli base +field-create --base-token "<base_token>" --table-id "<customers_table_id>" --json '{"type":"user","name":"负责人","multiple":false}' --as user
sleep 1.5
lark-cli base +field-create --base-token "<base_token>" --table-id "<customers_table_id>" --json '{"type":"select","name":"重要程度","multiple":false,"options":[{"name":"高"},{"name":"中"},{"name":"低"}]}' --as user
sleep 1.5
lark-cli base +field-create --base-token "<base_token>" --table-id "<customers_table_id>" --json '{"type":"select","name":"客户状态","multiple":false,"options":[{"name":"活跃"},{"name":"沉睡"},{"name":"流失"}]}' --as user
sleep 1.5
lark-cli base +field-create --base-token "<base_token>" --table-id "<customers_table_id>" --json '{"type":"text","name":"备注"}' --as user
sleep 2

Opportunity Management table fields

lark-cli base +field-create --base-token "<base_token>" --table-id "<opportunities_table_id>" --json '{"type":"text","name":"商机名称"}' --as user
sleep 1.5
lark-cli base +field-create --base-token "<base_token>" --table-id "<opportunities_table_id>" --json '{"type":"number","name":"商机金额","style":{"type":"currency","precision":2,"currency_code":"CNY"}}' --as user
sleep 1.5
lark-cli base +field-create --base-token "<base_token>" --table-id "<opportunities_table_id>" --json '{"type":"select","name":"商机阶段","multiple":false,"options":[{"name":"待接触"},{"name":"需求确认"},{"name":"方案报价"},{"name":"商务谈判"},{"name":"赢单"},{"name":"丢单"}]}' --as user
sleep 1.5
lark-cli base +field-create --base-token "<base_token>" --table-id "<opportunities_table_id>" --json '{"type":"user","name":"负责人","multiple":false}' --as user
sleep 1.5
lark-cli base +field-create --base-token "<base_token>" --table-id "<opportunities_table_id>" --json '{"type":"datetime","name":"预计成交日期","style":{"format":"yyyy-MM-dd"}}' --as user
sleep 1.5
lark-cli base +field-create --base-token "<base_token>" --table-id "<opportunities_table_id>" --json '{"type":"text","name":"竞争对手"}' --as user
sleep 1.5
lark-cli base +field-create --base-token "<base_token>" --table-id "<opportunities_table_id>" --json '{"type":"select","name":"丢单原因","multiple":false,"options":[{"name":"价格"},{"name":"功能"},{"name":"竞品"},{"name":"无预算"},{"name":"项目取消"},{"name":"其他"}]}' --as user
sleep 1.5
lark-cli base +field-create --base-token "<base_token>" --table-id "<opportunities_table_id>" --json '{"type":"select","name":"商机来源","multiple":false,"options":[{"name":"新开发"},{"name":"老客户增购"},{"name":"转介绍"}]}' --as user
sleep 2

# 添加系统字段：创建时间（公式字段"商机天数"依赖此字段）
lark-cli base +field-create --base-token "<base_token>" --table-id "<opportunities_table_id>" --json '{"type":"created_at","name":"创建时间","style":{"format":"yyyy-MM-dd HH:mm"}}' --as user

Follow-up Records table fields

lark-cli base +field-create --base-token "<base_token>" --table-id "<follow_ups_table_id>" --json '{"type":"select","name":"跟进方式","multiple":false,"options":[{"name":"电话"},{"name":"微信"},{"name":"面访"},{"name":"邮件"},{"name":"视频会议"},{"name":"其他"}]}' --as user
sleep 1.5
lark-cli base +field-create --base-token "<base_token>" --table-id "<follow_ups_table_id>" --json '{"type":"text","name":"跟进摘要"}' --as user
sleep 1.5
lark-cli base +field-create --base-token "<base_token>" --table-id "<follow_ups_table_id>" --json '{"type":"text","name":"客户反馈"}' --as user
sleep 1.5
lark-cli base +field-create --base-token "<base_token>" --table-id "<follow_ups_table_id>" --json '{"type":"text","name":"下一步计划"}' --as user
sleep 1.5
lark-cli base +field-create --base-token "<base_token>" --table-id "<follow_ups_table_id>" --json '{"type":"datetime","name":"下次跟进时间","style":{"format":"yyyy-MM-dd HH:mm"}}' --as user
sleep 1.5
lark-cli base +field-create --base-token "<base_token>" --table-id "<follow_ups_table_id>" --json '{"type":"user","name":"跟进人","multiple":false}' --as user
sleep 1.5
lark-cli base +field-create --base-token "<base_token>" --table-id "<follow_ups_table_id>" --json '{"type":"select","name":"商机阶段变更","multiple":false,"options":[{"name":"无变更"},{"name":"推进到下一阶段"},{"name":"退回上一阶段"},{"name":"赢单"},{"name":"丢单"}]}' --as user
sleep 1.5

# 添加系统字段：创建时间（公式字段"最后跟进时间"依赖此字段）
lark-cli base +field-create --base-token "<base_token>" --table-id "<follow_ups_table_id>" --json '{"type":"created_at","name":"创建时间","style":{"format":"yyyy-MM-dd HH:mm"}}' --as user

Step 1.5: Create bidirectional link fields (with a 2-second interval between each creation)

CRITICAL: link fields MUST use bidirectional: true, otherwise cross-table aggregation cannot be done in formulas/lookups.

# 商机管理表 → 客户信息表
lark-cli base +field-create --base-token "<base_token>" --table-id "<opportunities_table_id>" \
  --json '{"type":"link","name":"关联客户","link_table":"客户信息","bidirectional":true,"bidirectional_link_field_name":"关联商机"}' \
  --as user
sleep 2

# 跟进记录表 → 客户信息表
lark-cli base +field-create --base-token "<base_token>" --table-id "<follow_ups_table_id>" \
  --json '{"type":"link","name":"关联客户","link_table":"客户信息","bidirectional":true,"bidirectional_link_field_name":"跟进记录"}' \
  --as user
sleep 2

# 跟进记录表 → 商机管理表
lark-cli base +field-create --base-token "<base_token>" --table-id "<follow_ups_table_id>" \
  --json '{"type":"link","name":"关联商机","link_table":"商机管理","bidirectional":true,"bidirectional_link_field_name":"跟进记录"}' \
  --as user
sleep 2

Step 1.6: Create formula fields

IMPORTANT: before creating a formula field, you MUST first read ../lark-base/references/formula-field-guide.md, and after creation append --i-have-read-guide

Opportunity Management table: Suggested Win Probability

Automatically suggest the win probability based on the opportunity stage.

lark-cli base +field-create --base-token "<base_token>" --table-id "<opportunities_table_id>" \
  --json '{"type":"formula","name":"赢单概率建议","expression":"SWITCH([商机阶段], \"待接触\", 10, \"需求确认\", 30, \"方案报价\", 50, \"商务谈判\", 75, \"赢单\", 100, 0)"}' \
  --i-have-read-guide \
  --as user
sleep 2

Opportunity Management table: Opportunity Days

The number of days since the opportunity was created.

lark-cli base +field-create --base-token "<base_token>" --table-id "<opportunities_table_id>" \
  --json '{"type":"formula","name":"商机天数","expression":"DAYS(NOW(), [创建时间])"}' \
  --i-have-read-guide \
  --as user
sleep 2

Customer Info table: Customer Opportunity Count

Cross-table count via the link field.

lark-cli base +field-create --base-token "<base_token>" --table-id "<customers_table_id>" \
  --json '{"type":"formula","name":"商机数量","expression":"COUNTA([关联商机])"}' \
  --i-have-read-guide \
  --as user
sleep 2

Customer Info table: Customer Cumulative Won Amount

Cross-table conditional aggregation via a formula.

lark-cli base +field-create --base-token "<base_token>" --table-id "<customers_table_id>" \
  --json '{"type":"formula","name":"累计赢单金额","expression":"SUM([关联商机].[商机金额].FILTER([关联商机].[商机阶段] = \"赢单\"))"}' \
  --i-have-read-guide \
  --as user
sleep 2

Customer Info table: Last Follow-up Time

Take the most recent follow-up time from the follow-up records.

lark-cli base +field-create --base-token "<base_token>" --table-id "<customers_table_id>" \
  --json '{"type":"formula","name":"最后跟进时间","expression":"MAX([跟进记录].[创建时间])"}' \
  --i-have-read-guide \
  --as user
sleep 2

Note: the Follow-up Records table's 创建时间 is the created_at system field, referenced in formulas via the bidirectional link field [跟进记录] . Make sure the Follow-up Records table has created the created_at system field (already added in Step 1.4).

Opportunity Management table: Follow-up Count

Cross-table count via the link field.

lark-cli base +field-create --base-token "<base_token>" --table-id "<opportunities_table_id>" \
  --json '{"type":"formula","name":"跟进次数","expression":"COUNTA([跟进记录])"}' \
  --i-have-read-guide \
  --as user

If the formula uses the created_at system field, you need to use +table-get to get the actual system field name (it may be called "Created Time"), and make sure the field name in the formula matches the +field-list return value.

Step 1.7: Create views

View 1: Opportunity Stage Kanban

lark-cli base +view-create --base-token "<base_token>" --table-id "<opportunities_table_id>" \
  --json '{"name":"商机阶段看板","type":"kanban"}' \
  --as user
sleep 2

# 设置看板分组：按商机阶段分组
lark-cli base +view-set-group --base-token "<base_token>" --table-id "<opportunities_table_id>" --view-id "<kanban_view_id>" \
  --json '[{"field":"商机阶段"}]' \
  --as user

View 2: Follow-up Timeline

lark-cli base +view-create --base-token "<base_token>" --table-id "<follow_ups_table_id>" \
  --json '{"name":"跟进时间线","type":"grid"}' \
  --as user
sleep 2

# 设置排序：按创建时间倒序
lark-cli base +view-set-sort --base-token "<base_token>" --table-id "<follow_ups_table_id>" --view-id "<timeline_view_id>" \
  --json '[{"field_name":"创建时间","desc":true}]' \
  --as user

View 3: Won This Month

lark-cli base +view-create --base-token "<base_token>" --table-id "<opportunities_table_id>" \
  --json '{"name":"本月赢单","type":"grid"}' \
  --as user
sleep 2

# 设置筛选：商机阶段 = 赢单
lark-cli base +view-set-filter --base-token "<base_token>" --table-id "<opportunities_table_id>" --view-id "<won_view_id>" \
  --json '{"conjunction":"and","conditions":[{"field_name":"商机阶段","operator":"is","value":["赢单"]}]}' \
  --as user

Step 1.8: Create a dashboard

IMPORTANT: before creating a dashboard Block, first read ../lark-base/references/dashboard-block-data-config.md

# 创建仪表盘
lark-cli base +dashboard-create --base-token "<base_token>" --name "销售看板" --as user

Record the returned dashboard_id。

Block 1: Sales Funnel

lark-cli base +dashboard-block-create --base-token "<base_token>" --dashboard-id "<dashboard_id>" \
  --name "销售漏斗" --type funnel \
  --data-config '{"table_name":"商机管理","count_all":true,"group_by":[{"field_name":"商机阶段","mode":"integrated"}]}' \
  --as user
sleep 1.5

Block 2: Won Amount This Month

lark-cli base +dashboard-block-create --base-token "<base_token>" --dashboard-id "<dashboard_id>" \
  --name "本月赢单金额" --type statistics \
  --data-config '{"table_name":"商机管理","series":[{"field_name":"商机金额","rollup":"SUM"}],"filter":{"conjunction":"and","conditions":[{"field_name":"商机阶段","operator":"is","value":"赢单"}]}}' \
  --as user
sleep 1.5

Block 3: Total Opportunity Amount (In Progress)

lark-cli base +dashboard-block-create --base-token "<base_token>" --dashboard-id "<dashboard_id>" \
  --name "商机总金额" --type statistics \
  --data-config '{"table_name":"商机管理","series":[{"field_name":"商机金额","rollup":"SUM"}],"filter":{"conjunction":"and","conditions":[{"field_name":"商机阶段","operator":"isNot","value":"赢单"},{"field_name":"商机阶段","operator":"isNot","value":"丢单"}]}}' \
  --as user
sleep 1.5

Block 4: Customer Count

lark-cli base +dashboard-block-create --base-token "<base_token>" --dashboard-id "<dashboard_id>" \
  --name "客户总数" --type statistics \
  --data-config '{"table_name":"客户信息","count_all":true}' \
  --as user
sleep 1.5

Block 5: Stage Distribution

lark-cli base +dashboard-block-create --base-token "<base_token>" --dashboard-id "<dashboard_id>" \
  --name "阶段分布" --type pie \
  --data-config '{"table_name":"商机管理","count_all":true,"group_by":[{"field_name":"商机阶段","mode":"integrated"}]}' \
  --as user
sleep 1.5

Block 6: Owner Performance Ranking

lark-cli base +dashboard-block-create --base-token "<base_token>" --dashboard-id "<dashboard_id>" \
  --name "负责人业绩排行" --type column \
  --data-config '{"table_name":"商机管理","series":[{"field_name":"商机金额","rollup":"SUM"}],"group_by":[{"field_name":"负责人","mode":"integrated"}],"filter":{"conjunction":"and","conditions":[{"field_name":"商机阶段","operator":"is","value":"赢单"}]}}' \
  --as user
sleep 1.5

Step 1.9: Create a form

Create a customer inquiry registration form on the Customer Info table so customers can submit information themselves.

lark-cli base +form-create --base-token "<base_token>" --table-id "<customers_table_id>" \
  --name "客户咨询登记" \
  --description "欢迎填写咨询信息，我们会尽快与您联系" \
  --as user

After the form is created, you can use +form-questions-list to view existing questions, then use +form-questions-create to add or adjust questions.

Phase 2: Customer Management

2.1 Add Customer

lark-cli base +record-upsert --base-token "<base_token>" --table-id "<customers_table_id>" \
  --json '{"客户名称":"北京星辰科技有限公司","客户类型":"企业客户","联系人":"王总","手机号":"138xxxx","邮箱":"wang@xx.com","行业":"互联网","公司规模":"51-200人","来源渠道":"展会","负责人":[{"id":"ou_xxx"}],"重要程度":"高","备注":""}' \
  --as user

Agent processing logic：

Parse user input: extract customer name, customer type, contact, phone, email, industry, size, source, owner, importance, notes
Leave fields not provided empty
Industry/company size/source channel must match existing options; if not among the options, prompt the user
Personnel fields require user_id, which can be found via contact +search-user search

New customer reply format：

客户创建成功！

客户名称：北京星辰科技有限公司
客户类型：企业客户
联系人：王总
手机号：138xxxx
行业：互联网
负责人：当前用户

2.2 Modify customer information

# 先查询获取 record_id
lark-cli base +record-list --base-token "<base_token>" --table-id "<customers_table_id>" --as user

# 更新客户信息（Delta 更新，只更新用户指定的字段）
lark-cli base +record-upsert --base-token "<base_token>" --table-id "<customers_table_id>" \
  --record-id "<record_id>" \
  --json '{"重要程度":"高","备注":"重点客户"}' \
  --as user

2.3 Query customer list

lark-cli base +record-list --base-token "<base_token>" --table-id "<customers_table_id>" \
  --limit 200 --as user

Agent processing：

Filter by user-specified conditions (industry, owner, importance level, etc.)
Organize into an easy-to-read table format

2.4 Delete customer

lark-cli base +record-delete --base-token "<base_token>" --table-id "<customers_table_id>" \
  --record-id "<record_id>" --yes --as user

Phase 3: Opportunity management

3.1 Create opportunity

lark-cli base +record-upsert --base-token "<base_token>" --table-id "<opportunities_table_id>" \
  --json '{"商机名称":"企业IM系统采购","关联客户":[{"id":"<customer_record_id>"}],"商机金额":500000,"商机阶段":"待接触","负责人":[{"id":"ou_xxx"}],"预计成交日期":"2026-06-30","竞争对手":"","商机来源":"新开发"}' \
  --as user

Related field value format: The value of the related field is in [{id: "recXXXX"}] object array format. Date format: Use the YYYY-MM-DD HH:mm:ss string format.

Agent processing logic：

Parse user input: extract opportunity name, related customer, amount, stage, owner, expected closing date, etc.
If the user provides a customer name, first look up the corresponding record_id in the customer table
If the customer does not exist, ask whether to also create a customer record
The default stage is "Pending Contact" and the default opportunity source is "New Development"

Create opportunity reply format：

商机创建成功！

商机名称：企业IM系统采购
关联客户：北京星辰科技
商机金额：¥500,000.00
商机阶段：待接触
负责人：当前用户
预计成交日期：2026-06-30

3.2 Advance opportunity stage

When the user mentions a stage change in a follow-up record, or directly asks to advance the opportunity:

# 更新商机阶段
lark-cli base +record-upsert --base-token "<base_token>" --table-id "<opportunities_table_id>" \
  --record-id "<opportunity_record_id>" \
  --json '{"商机阶段":"方案报价"}' \
  --as user

Agent processing logic：

User says "advance to Proposal Quotation" → update the opportunity stage to "Proposal Quotation"
User says "won the deal" → update the opportunity stage to "Won"
User says "lost the deal" → update the opportunity stage to "Lost" and ask for the reason for the loss
After a stage change, automatically update the win probability suggestion (formula field is calculated automatically)

Stage order: Pending Contact → Needs Confirmation → Proposal Quotation → Business Negotiation → Won/Lost

3.3 Query opportunity list

lark-cli base +record-list --base-token "<base_token>" --table-id "<opportunities_table_id>" \
  --limit 200 --as user

Agent processing：

Filter by conditions such as stage, owner, customer, etc.
Sort by opportunity amount

3.4 Sales funnel query

lark-cli base +data-query --base-token "<base_token>" \
  --dsl '{
    "datasource": {"type": "table", "table": {"tableId": "<opportunities_table_id>"}},
    "dimensions": [{"field_name":"商机阶段","alias":"stage"}],
    "measures": [
      {"field_name":"商机金额","aggregation":"sum","alias":"total_amount"},
      {"field_name":"商机名称","aggregation":"count","alias":"count"}
    ]
  }' \
  --as user

Reply format：

销售漏斗

| 阶段 | 数量 | 金额 |
|------|------|------|
| 待接触 | 12 | ¥1,200,000 |
| 需求确认 | 8 | ¥800,000 |
| 方案报价 | 5 | ¥500,000 |
| 商务谈判 | 3 | ¥450,000 |
| 赢单 | 2 | ¥200,000 |
| 丢单 | 1 | ¥100,000 |

Phase 4: Follow-up records

4.1 Add follow-up record

lark-cli base +record-upsert --base-token "<base_token>" --table-id "<follow_ups_table_id>" \
  --json '{"关联客户":[{"id":"<customer_record_id>"}],"关联商机":[{"id":"<opportunity_record_id>"}],"跟进方式":"电话","跟进摘要":"电话联系了王总，对方对方案比较满意","客户反馈":"认可方案，但价格偏高","下一步计划":"下周三面访确认细节","下次跟进时间":"2026-04-16 14:00","跟进人":[{"id":"ou_xxx"}],"商机阶段变更":"无变更"}' \
  --as user

Agent processing logic：

Parse user input: extract follow-up method, summary, customer feedback, next steps, next follow-up time, stage change
Look up the record_id of the related customer and opportunity
Key: If the user mentions a stage change (such as "the other party agrees to advance to Proposal Quotation"), synchronously update the opportunity stage of the corresponding opportunity
Date defaults to the current day

Follow-up record reply format：

跟进记录已保存！

客户：北京星辰科技
商机：企业IM系统采购
跟进方式：电话
跟进摘要：电话联系了王总，对方对方案比较满意
下一步计划：下周三面访确认细节
下次跟进时间：2026-04-16 14:00

4.2 Query follow-up records

lark-cli base +record-list --base-token "<base_token>" --table-id "<follow_ups_table_id>" \
  --limit 200 --as user

4.3 Sales daily report

Agent 处理逻辑：
1. 查询今天所有跟进记录（按创建时间筛选当天）
2. 查询今天新增的客户（按创建时间筛选当天）
3. 查询今天新增的商机
4. 查询今天发生阶段变更的商机
5. 生成结构化日报

Daily report format：

销售日报 — 2026-04-13

一、今日跟进（3 条）
| 客户 | 商机 | 方式 | 摘要 | 阶段变更 |
|------|------|------|------|---------|
| 北京星辰 | IM采购 | 电话 | 方案满意 | → 方案报价 |
| 上海悦达 | 数据服务 | 微信 | 初步接触 | 无 |

二、新增客户（1 家）
- 深圳华创科技，互联网行业，来自线上广告

三、新增商机（2 个）
- 深圳华创科技：云服务采购 ¥300,000（待接触）
- 北京星辰科技：IM系统采购 ¥500,000（待接触）

四、阶段变更（1 个）
- 北京星辰 IM采购：待接触 → 方案报价

五、今日提醒
- 1 个商机超 7 天未跟进
- 2 个客户待跟进

Phase 5: Data analysis

5.1 Performance ranking

lark-cli base +data-query --base-token "<base_token>" \
  --dsl '{
    "datasource": {"type": "table", "table": {"tableId": "<opportunities_table_id>"}},
    "dimensions": [{"field_name":"负责人","alias":"owner"}],
    "measures": [
      {"field_name":"商机金额","aggregation":"sum","alias":"total_amount"},
      {"field_name":"商机名称","aggregation":"count","alias":"count"}
    ],
    "filters": {"type": 1, "conjunction": "and", "conditions": [{"field_name":"商机阶段","operator":"is","value": ["赢单"]}]}
  }' \
  --as user

5.2 This month's won deal statistics

lark-cli base +data-query --base-token "<base_token>" \
  --dsl '{
    "datasource": {"type": "table", "table": {"tableId": "<opportunities_table_id>"}},
    "measures": [
      {"field_name":"商机金额","aggregation":"sum","alias":"total_amount"},
      {"field_name":"商机名称","aggregation":"count","alias":"count"}
    ],
    "filters": {"type": 1, "conjunction": "and", "conditions": [{"field_name":"商机阶段","operator":"is","value": ["赢单"]}]}
  }' \
  --as user

5.3 Opportunity conversion rate

# 先获取各阶段数量
lark-cli base +data-query --base-token "<base_token>" \
  --dsl '{
    "datasource": {"type": "table", "table": {"tableId": "<opportunities_table_id>"}},
    "dimensions": [{"field_name":"商机阶段","alias":"stage"}],
    "measures": [{"field_name":"商机名称","aggregation":"count","alias":"count"}]
  }' \
  --as user

Agent calculates the conversion rate for each stage:

商机转化分析

| 阶段 | 数量 | 转化率 |
|------|------|--------|
| 待接触 | 12 | 100%（入口） |
| 需求确认 | 8 | 67% |
| 方案报价 | 5 | 42% |
| 商务谈判 | 3 | 25% |
| 赢单 | 2 | 17% |

整体赢单率：17%（2/12）

Phase 6: Automated reminders

6.1 Daily follow-up reminder

Query the content that needs follow-up today:

# 查询下次跟进时间 = 今天的跟进记录
lark-cli base +record-list --base-token "<base_token>" --table-id "<follow_ups_table_id>" \
  --limit 200 --as user
# AI 筛选：下次跟进时间 = 今天

# 查询预计成交日期 = 今天且未关单的商机
lark-cli base +record-list --base-token "<base_token>" --table-id "<opportunities_table_id>" \
  --limit 200 --as user
# AI 筛选：预计成交日期 = 今天 AND 商机阶段 NOT IN (赢单, 丢单)

Send Feishu message:

lark-cli im +messages-send --chat-id "<chat_id>" \
  --msg-type "interactive" \
  --content '{
    "config": {"wide_screen_mode": true},
    "header": {"title": {"tag": "plain_text", "content": "今日跟进提醒"}, "template": "blue"},
    "elements": [
      {"tag": "div", "text": {"tag": "lark_md", "content": "**跟进提醒**\n- 客户A：下次跟进时间到，计划：XXX\n\n**商机提醒**\n- 商机B：预计今天成交，金额：XXX"}}
    ]
  }' \
  --as user

Important: Before sending a message, first read ../lark-im/SKILL.md to understand the message sending rules.

6.2 Overdue follow-up alert

# 查询超过 7 天未跟进的商机
lark-cli base +record-list --base-token "<base_token>" --table-id "<opportunities_table_id>" \
  --limit 200 --as user
# AI 筛选：商机阶段 NOT IN (赢单, 丢单) AND 跟进次数 > 0 AND 最后跟进时间 > 7 天前

Message format：

超时未跟进预警

以下商机已超过 7 天未跟进：

| 商机 | 客户 | 阶段 | 未跟进天数 |
|------|------|------|-----------|
| 商机A | 客户A | 方案报价 | 8 天 |
| 商机B | 客户B | 需求确认 | 12 天 |

6.3 Won deal celebration

When the opportunity stage changes to "Won", send a Feishu group message notification:

lark-cli im +messages-send --chat-id "<chat_id>" \
  --msg-type "interactive" \
  --content '{
    "config": {"wide_screen_mode": true},
    "header": {"title": {"tag": "plain_text", "content": "赢单喜报！"}, "template": "green"},
    "elements": [
      {"tag": "div", "text": {"tag": "lark_md", "content": "商机：XX公司 ERP 采购\n金额：¥100,000\n负责人：@用户\n商机周期：XX天"}}
    ]
  }' \
  --as user

6.4 Weekly Sales Briefing

Agent 处理逻辑：
1. 汇总上周新增客户数（按创建时间筛选）
2. 汇总上周新增商机数和金额
3. 汇总上周赢单数和金额
4. 汇总当前漏斗各阶段数量和金额
5. 生成结构化简报并发送

Briefing Format：

周销售简报（上周 2026-04-07 ~ 2026-04-13）

一、关键指标
- 新增客户：5 家
- 新增商机：8 个，总金额 ¥500,000
- 赢单：3 个，金额 ¥200,000
- 丢单：1 个，金额 ¥50,000

二、当前漏斗
| 阶段 | 数量 | 金额 |
|------|------|------|
| 待接触 | 12 | ¥1,200,000 |
| 需求确认 | 8 | ¥800,000 |
| 方案报价 | 5 | ¥500,000 |
| 商务谈判 | 3 | ¥450,000 |

三、需关注
- 3 个商机超 7 天未跟进
- 2 个商机预计本周成交

Phase 7: Form Management

7.1 Create a Customer Inquiry Registration Form

lark-cli base +form-create --base-token "<base_token>" --table-id "<customers_table_id>" \
  --name "客户咨询登记" \
  --description "欢迎填写咨询信息，我们会尽快与您联系" \
  --as user

7.2 View and Configure Form Questions

# 查看已有问题
lark-cli base +form-questions-list --base-token "<base_token>" --table-id "<customers_table_id>" --form-id "<form_id>" --as user

# 添加/修改问题
lark-cli base +form-questions-create --base-token "<base_token>" --table-id "<customers_table_id>" --form-id "<form_id>" \
  --json '[{"name":"客户名称","type":"input"},{"name":"联系电话","type":"input"},{"name":"咨询内容","type":"textarea"}]' \
  --as user

Intent → Command Index

Intent	Recommended Command	Remarks
Create CRM System	`+base-create` → 3 × `+table-create` → one by one `+field-create` → bidirectional link → formula → view → dashboard → form	Execute in the order of Step 1
Add Customer	`+record-upsert` → customers table
Update Customer	`+record-list` → `+record-upsert` → customers table	Look up record_id first
Delete Customer	`+record-delete` → customers table	with `--yes`
Query Customer	`+record-list` → customers table	AI filtering
Create Opportunity	`+record-list`(look up customer) → `+record-upsert` → opportunities table	Link Customer
Advance Opportunity	`+record-upsert` → opportunities table	Update Opportunity Stage
View Opportunity	`+record-list` → opportunities table
Sales Funnel	`+data-query` → opportunities table	Aggregate by stage
Add Follow-up	`+record-upsert` → follow_ups table	If the stage changes, update the opportunity accordingly
Sales Daily Report	`+record-list` → follow_ups + customers + opportunities tables	AI summary
Performance Ranking	`+data-query` → opportunities table	Aggregate won amount by owner
Follow-up Reminder	`+record-list` → follow_ups + opportunities → `im +messages-send`
Overdue Alert	`+record-list` → opportunities → `im +messages-send`
Won Deal Announcement	`im +messages-send`
Weekly Briefing	`+record-list` + `+data-query` → `im +messages-send`
Create Form	`+form-create` → `+form-questions-create`
Create Dashboard	`+dashboard-create` → `+dashboard-block-create`	Read the dashboard guide first

Common Errors Quick Reference

Error Scenario	Cause	Solution
`+table-create --fields` Rate limit 800004135	Creating multiple fields at once triggers rate limiting	Create an empty table first, then use `+field-create` to add fields one by one with a 1.5-second interval
Formula field cross-table query returns results for the entire table	Link field is not bidirectional	Link fields must be `bidirectional: true`
A unidirectional link cannot be changed to bidirectional	API restriction unsafe_operation_blocked	Delete the old link field and recreate a bidirectional link
`+data-query` unknown flag `--table-id`	This command does not support `--table-id`	Put table-id in the `--dsl` of `datasource.table.tableId` in
`+data-query` does not support aggregation on a certain field	Formula/lookup/link fields are not in the whitelist	Only aggregate stored fields
`+data-query` aggregation `counta` not supported	`+data-query` only supports `count` does not support `counta`	Use instead `"aggregation":"count"`
Failed to write linked field value	Format error	Linked field value must be `[{id: "recXXX"}]` object array
Failed to write date field	Format error	Use `YYYY-MM-DD HH:mm:ss` string format
Failed to write person field	Format error	Use `[{id: "ou_xxx"}]` object array
Failed to create formula field	Guide not read or expression is incorrect	First read formula-field-guide.md
Table name/field name mismatch	Guessing from natural language	Use `+table-list` / `+field-list` to get the real name
Concurrency error 1254291	list command concurrency	All `+xxx-list` executed serially
Failed to create dashboard Block	Use `--json` instead of standalone parameters	Switch to `--name` + `--type` + `--data-config` standalone parameters
`+view-create` does not support `--name`	Must use `--json` to pass parameters	Switch to `--json '{"name":"xxx","type":"grid"}'`
`+view-set-group` field name key error	JSON key is `field` not `field_name`	Switch to `[{"field":"字段名"}]`
Formula referencing "Created time" returns null	Table has no `created_at` system field	During initialization, use `+field-create` to manually add `type:"created_at"` field
Base comes with default "Data table"	Automatically generated when creating a Base	Ignore this table, and only operate on the three tables: Customer Information/Opportunity Management/Follow-up Records
Failed to send message	chat_id is invalid	Use `im +chat-list` to confirm chat_id
Insufficient permissions	scope not enabled	Refer to lark-shared to handle permission issues

Permission table

Command	Required scope
`base +base-create`	`bitable:app`
`base +table-create`	`bitable:app`
`base +field-create`	`bitable:app`
`base +field-list`	`bitable:app:readonly`
`base +record-list`	`bitable:app:readonly`
`base +record-upsert`	`bitable:app`
`base +record-delete`	`bitable:app`
`base +data-query`	`bitable:app:readonly`
`base +view-create`	`bitable:app`
`base +view-set-filter`	`bitable:app`
`base +view-set-sort`	`bitable:app`
`base +dashboard-create`	`bitable:app`
`base +dashboard-block-create`	`bitable:app`
`base +form-create`	`bitable:app`
`base +form-questions-list`	`bitable:app:readonly`
`base +form-questions-create`	`bitable:app`
`im +messages-send`	`im:message:send_as_bot`
`contact +search-user`	`contact:employee.id:readonly`

Reference

../lark-shared/SKILL.md — Authentication, permissions (required reading)
../lark-base/SKILL.md — Bitable atomic operations
../lark-base/references/formula-field-guide.md — Formula field syntax (required reading before creating formula fields)
../lark-base/references/lookup-field-guide.md — lookup field configuration rules
../lark-base/references/lark-base-shortcut-field-properties.md — Field property JSON specification
../lark-base/references/lark-base-shortcut-record-value.md — Record value format specification
../lark-base/references/lark-base-data-query.md — Aggregation analysis DSL
../lark-base/references/dashboard-block-data-config.md — Dashboard Block configuration
../lark-base/references/lark-base-form-create.md — Form creation
../lark-base/references/lark-base-view-set-filter.md — View filter configuration
../lark-im/SKILL.md — Feishu message sending
../lark-contact/SKILL.md — Contacts person query

About this skill

Applicable Scenarios

Prerequisites

System Architecture

Agent Core Rules

Phase 1: System Initialization

Step 1.1: Create Base

Step 1.2: Create data tables (create empty tables first, with a 2-second interval)

Step 1.3: Get the table_id of each table

Step 1.4: Add storage fields one by one (with a 1.5-second interval between each creation)

Customer Info table fields

Opportunity Management table fields

Follow-up Records table fields

Step 1.5: Create bidirectional link fields (with a 2-second interval between each creation)

Step 1.6: Create formula fields

Opportunity Management table: Suggested Win Probability

Opportunity Management table: Opportunity Days

Customer Info table: Customer Opportunity Count

Customer Info table: Customer Cumulative Won Amount

Customer Info table: Last Follow-up Time

Opportunity Management table: Follow-up Count

Step 1.7: Create views

View 1: Opportunity Stage Kanban

View 2: Follow-up Timeline

View 3: Won This Month

Step 1.8: Create a dashboard

Block 1: Sales Funnel

Block 2: Won Amount This Month

Block 3: Total Opportunity Amount (In Progress)

Block 4: Customer Count

Block 5: Stage Distribution

Block 6: Owner Performance Ranking

Step 1.9: Create a form

Phase 2: Customer Management

2.1 Add Customer

2.2 Modify customer information

2.3 Query customer list

2.4 Delete customer

Phase 3: Opportunity management

3.1 Create opportunity

3.2 Advance opportunity stage

3.3 Query opportunity list

3.4 Sales funnel query

Phase 4: Follow-up records

4.1 Add follow-up record

4.2 Query follow-up records

4.3 Sales daily report

Phase 5: Data analysis

5.1 Performance ranking

5.2 This month's won deal statistics

5.3 Opportunity conversion rate

Phase 6: Automated reminders

6.1 Daily follow-up reminder

6.2 Overdue follow-up alert

6.3 Won deal celebration

6.4 Weekly Sales Briefing

Phase 7: Form Management

7.1 Create a Customer Inquiry Registration Form

7.2 View and Configure Form Questions

Intent → Command Index

Common Errors Quick Reference

Permission table

Reference