Case Study: Auto-Create Requirements and Send Notifications with Claude Code + Lovrabet + Lark CLI
Scenario
With Lovrabet CLI installed, just say one sentence to Claude Code — for example, "Create a requirement for Zixuan: add the page editing guide to the help center ASAP" — and Claude Code will automatically:
- Identify the assignee
- Create the requirement in the Lovrabet system
- Send an email notification to the assignee
No manual steps required.
Prerequisites
1. A Lovrabet Account
If you don't have one yet, register at the Lovrabet User Center.
2. An App with "Requirement Management" Enabled
You need a Lovrabet app that has completed database reverse analysis. This guide uses Qizhi Yuntu Project Management as an example (appcode: app-173e8652).
If you're using your own app, replace
app-173e8652and the related dataset codes with your own.
3. A Mac or Linux Computer
All steps have been verified on macOS. Linux users can follow the same commands.
Step 1: Install the CLI Tools
Open your terminal and run the following commands:
1.1 Install Lovrabet CLI and Skills
npm install -g lovrabet
npx skill add lovrabet/lovrabet-cli -g -y
1.2 Install lark-cli (Lark CLI)
npx @larksuite/cli@latest install
1.3 Install Claude Code or Another Agent Client
Install your preferred Agent IDE following its official documentation. This guide uses Claude Code as the example.
After installation, verify everything is in place:
lovrabet --version
lark-cli --version
claude-code --version
Step 2: Log In and Authorize
2.1 Log In to Lovrabet
Run in your terminal:
lovrabet auth login
Follow the prompts to enter your Lovrabet account credentials.
2.2 Verify App Access
# List registered apps
lovrabet app list --format json
What is an appcode? Every Lovrabet app has a unique identifier called an appcode — think of it as the app's "name." You need to tell the CLI which app to work with.
2.3 Log In to Lark
lark-cli auth login
This will open a browser. Follow the prompts to authorize with your Lark account. Return to the terminal once authorization is complete.
Step 3: Start Using
3.1 Launch Claude Code
Open a terminal in any directory and type:
claude
3.2 Submit a Request
Just say one sentence and Claude Code will handle the entire flow. For example:
/lovrabet Create a requirement for Zixuan: add the page editing guide to the help center and configure the entry point.
Additional context:
1) Look up the assignee's info from the members table
2) Match the project by title keyword
3) Write the requirement to the corresponding project management system
4) Send an email to the assignee — include the requirement description and a clickable link to the requirement detail page
5) Return the requirement ID and email delivery status when done
3.3 Wait for Execution
Claude Code will automatically:
- Query Zixuan's project member info
- Create the requirement (priority: high, status: new)
- Send an email notification to Zixuan
Once complete, you'll see the requirement ID and email delivery result.
Step 4: Save the Flow as a Reusable Skill (Optional)
If your team frequently creates requirements, having the AI figure it out from scratch each time is slow and inconsistent. You can save a working flow as a reusable Skill.
Why Save as a Skill?
- You notice a business flow is repeated often
- You want to codify internal expertise so the AI follows it instead of re-interpreting each time
- You want teammates to trigger the same flow with one sentence, without understanding the underlying steps
How to Do It (Requirement Creation Example)
After completing a full flow in Claude Code, just say:
Use the skill-creator skill to package the above operations into a business Skill for future reuse
The AI will analyze what it just did, which tools and parameters it used, and call skill-creator to generate a Skill file. It will tell you the Skill name and where it's saved.
Let's say it generates a Skill named project-issue.
How to Use It After Creation
From now on, just say:
Use "project-issue" to create a requirement for Zixuan: the app background color should automatically adapt to the user's phone case color
The AI will load the Skill automatically and follow its steps — no need to re-describe the flow.
FAQ
Q: The claude command doesn't respond?
Make sure Claude Code CLI is installed:
claude-code --version
If you get "command not found," reinstall:
npm install -g @anthropic-ai/claude-code
Q: lovrabet auth login fails?
Make sure you're using a registered Lovrabet account and that your AccessKey is correct.
Q: lark-cli auth login shows a permission error?
Your Lark account may not have the required app permissions. Contact your Lark admin to grant access to the relevant Lark app.
Q: How do I view created requirements?
Check them inside the project management app you created in Lovrabet. The CLI is designed to let Agents operate Lovrabet-built apps — simulating human actions — so all results live within the app.
Technical Architecture (Optional Reading)
This solution uses a three-layer architecture:
| Layer | Tool | Responsibility |
|---|---|---|
| Orchestration | Claude Code + Global Skills | Understand intent, orchestrate flow, map members |
| Data | Lovrabet CLI | App switching, data CRUD |
| Outreach | lark-cli | Email notification delivery |
Each layer has its own responsibility. The Skill binds them into a complete business scenario.