Meegle CLI: manage Feishu Project work items and schedules in the terminal

Who it is for

If you already use Meegle (Feishu Project) and want to manage work items, schedules, and data in the terminal without opening a browser, this tool fits. It is the Meegle command-line tool. The explainer page is

https://github.com/larksuite/meegle-cli

A better fit already has a Meegle space you can sign into, local Node.js no lower than 16 (with npm / npx included), and accepts only this official install command. The install wizard will: install or upgrade the CLI globally, install the bundled AI Agent Skill, configure the Meegle host, and start sign-in. All four steps change the local environment. It fits people willing to finish authorization on this machine.

If you only want to click work items in the web UI and do not plan to install a global CLI or let an agent call project data, you do not need this wizard. Official text says the wizard installs the bundled skill into several coding environments and also mentions other agents; do not read the named ones as an exclusive roster, and do not hunt an undocumented marketplace install. The first pass should be read-only viewing, not creating or changing a schedule. The wizard binds the CLI and the skill in the same install; you cannot copy a skill folder to skip sign-in. Without a Meegle host address and an account that can sign in, prepare those two first, then run npx.

First try in 3 minutes

Use only this official install path:

npx @lark-project/meegle@latest install

Do not rewrite it as skills add, a plugin, or a marketplace instruction. Before you run it, confirm: local Node.js >= 16; you are ready to install the CLI globally on this machine; you are willing to configure the host next and finish sign-in. These four things have side effects. First look at where the wizard will write and which host you will sign into, then continue only after you confirm.

After install, restart the agent you will use and confirm the bundled skill has appeared. Official text does not write the skill slash command or a CLI subcommand inventory. The first trigger should use the skill name the wizard installed.

In an agent that is signed in and can see the skill, do one low-risk task: ask it to read-only view some work items or the current schedule. Have it first name the command or skill it will call, whether it writes locally, and whether it will change fields on Meegle. After you confirm it is pure read, allow the query.

Success signal: work-item or schedule data from Meegle appears in the terminal or the chat, and it no longer asks you to open a browser. If the wizard stops on a sign-in page, finish host setup and sign-in; do not switch to an undocumented package name. If the skill does not appear, restart the agent and confirm the wizard reached the skill-install step. If it still requires a browser to read data, sign-in or the host is unfinished; return to the wizard instead of installing another package.

Usage examples

Example 1: View work items in the terminal. Official capability is managing work items, schedules, and data. The first pass is view only: name a space or view you can see, and let the bundled skill or CLI read work-item titles and status. The page does not give a concrete subcommand spelling. If that is missing, ask the installed skill to “read-only show work items”; do not invent meegle xxx. Any create, assign, or close-ticket action stops at the plan step.

Example 2: View schedules and data, still without changing the browser flow. The same sentence places schedules, data, and work items side by side. The method is to read schedule or field data from the same space to check a calendar, not to change dates. Do not run if sign-in is unfinished. Official text does not write an export format or a bulk date-change interface; do not do undocumented bulk updates. Both examples assume the wizard already installed the CLI, the skill, and the host and finished sign-in; this is not a second install. If the chat shows a subcommand that never appears on the explainer page, first check it against the skill notes the wizard wrote; if it does not match, do not run it.

Limits

Runtime need is Node.js >= 16, with npm / npx. Below that version, do not force-change the install command. The wizard globally installs or upgrades the CLI, writes the skill, configures the host, and starts sign-in; you cannot “install the skill, skip sign-in” and still expect space data.

Official install is only npx @lark-project/meegle@latest install. Do not invent a second skills add or plugin-marketplace path. Environment names mentioned in the intro are not the only usable roster; unnamed environments are neither guaranteed nor ruled out. Trust the path the wizard actually writes.

Managing work items and schedules will hit write operations. Until you confirm, only allow reads. Official text has no subcommand table, and no sign-out, uninstall, or multi-host switch steps; if those are missing, stay inside the range the wizard already finished. Do not treat the product homepage or a Feishu Project intro page as a second install address. A global CLI upgrade may overwrite an older version; confirm you accept the upgrade before running the wizard. Sign-in opens an authorization flow; do not run it on a machine you do not trust.