DEV Community

Cover image for Feishu CLI Hands-on: Letting Codex Enter a Real Office Workflow
member_0af6418a
member_0af6418a

Posted on • Originally published at kunpeng-ai.com

Feishu CLI Hands-on: Letting Codex Enter a Real Office Workflow

#programming #ai #opensource #github

Feishu now has an official CLI, and I wanted to test a practical question:

Can an AI agent use it to enter a real office workflow?

I did not start by manually reading the documentation and turning it into a scripted demo. Instead, I gave the official larksuite/cli repository to Codex and asked it to figure out what the tool could do, install it, go through the configuration path, wait for human authorization, and then send me a message through Feishu.

After that, I turned the test into a small recurring workflow: a daily reminder to check SEO and GEO status for our main blog.

Feishu CLI hands-on

Why this is more than a command-line tool

Most CLI tools are useful because they turn repeated clicks into commands.

Feishu CLI is more interesting because the official README explicitly treats humans and AI agents as users of the tool. In the README I checked on June 1, 2026, the project describes support for messaging, docs, Bitable, spreadsheets, slides, calendar, mail, tasks, meetings, Markdown, and more.

It also describes 200+ commands and 26 AI Agent Skills.

That matters because agent-based office automation often gets stuck at the same point:

  • the agent can understand the task;
  • it can generate a plan or script;
  • but it has no stable, authorized way to operate the office system.

When office actions become available through a CLI, the agent can move from "understanding" to "executing a bounded action."

This does not mean full autonomous office work. It means repeated, low-risk, verifiable actions can start becoming workflows.

Feishu CLI capability surface

Letting Codex run the setup path

The first step was to let Codex read the official repository.

It found the larksuite/cli GitHub repo, read the Chinese README, and noticed that the documentation includes a quick-start path specifically for AI agents.

The flow looked like this:

  1. Codex confirmed the install command from the official README.
  2. It checked the local Node / npm environment.
  3. It ran the CLI install command.
  4. It entered the configuration flow.
  5. It returned authorization links to me.
  6. I completed authorization in the browser.
  7. Codex resumed and checked the authorization state.

The important part is the responsibility split.

The agent can read documentation, run commands, parse output, and prepare the next step. But authorization should stay human-controlled. Once an agent can operate an office system, permissions become a product and security question, not just a convenience feature.

Agent setup and authorization flow

The first useful loop: a Feishu message comes back

After installation and authorization, I tested the smallest complete loop:

human intent
-> agent understands the task
-> agent calls the CLI
-> CLI operates Feishu
-> Feishu message returns to the human
Enter fullscreen mode Exit fullscreen mode

That is small, but it is enough to prove the basic workflow path.

At that point, the question changes from "can this send a message?" to "what repeated office action should be turned into a workflow?"

I chose a daily SEO / GEO reminder.

The reminder is not complex. It asks me to check:

  • Google / Bing index status
  • search query and click changes
  • whether AI search or large models mention the brand
  • Chinese and English article titles, summaries, and links
  • whether recent content distribution created new entry points

This is exactly the kind of work that is important but easy to forget. A stable private reminder is more useful than a flashy automation that is too risky to run every day.

Start with small private tasks

The official README also includes a safety warning around AI-agent automation: hallucination, uncontrolled execution, and prompt injection are real risks when an agent operates an office platform under a user's authorization.

That should shape the first workflows.

My preferred starting point:

  • private reminders, not group-wide bots;
  • personal todos, not cross-team approvals;
  • fixed checklists, not open-ended execution;
  • read-only checks before write or delete actions;
  • no secrets, tokens, chat IDs, or open IDs in public screenshots, articles, or logs.

Feishu CLI safety boundary

Agent office automation should not begin with broad permissions. It should begin with low-risk, high-repeat, verifiable actions.

What this means for practical users

Many people still use AI mostly for Q&A, writing, or generating spreadsheet formulas.

Those are useful, but the bigger shift happens when agents can enter real workflows.

Feishu CLI is a good example. Once an office platform has a standardized command interface, an agent can help with:

  • daily metric reminders;
  • meeting follow-up summaries;
  • document summaries;
  • calendar conflict checks;
  • repeated spreadsheet updates;
  • fixed operational checklists.

None of these tasks are dramatic. But they are repeated, easy to forget, and valuable when they happen consistently.

The value of Feishu CLI is not that a command line can replace the Feishu client. Its value is that it gives agents an office-system entry point that can be installed, authorized, checked, executed, and interrupted by a human when needed.

Full write-up:

https://kunpeng-ai.com/en/blog/feishu-cli-ai-agent-workflow/

Top comments (0)