Skip to content
Get packs

Guide · Agent skill

How to use the Lark Native OpenAPI Explorer agent skill

Updated · written with AI from the project's own documentation and checked against our catalogue data

The lark-openapi-explorer skill is a set of instructions that lets an AI agent look up a native Feishu or Lark OpenAPI endpoint in the official documentation and call it directly with the lark-cli api command. It is meant for workspace tasks that existing lark-* skills and lark-cli commands do not cover, such as adding members to a group chat. This guide explains what the skill does, how to install it, how to trigger it, and what to check before letting it make changes.

Lark Native OpenAPI Explorerlarksuite/cli · MIT

Quick start

  1. 1Install lark-cli. Install lark-cli and confirm it is available on the command line, since the skill requires it. The source does not give the installation steps, so use the lark-cli documentation.
  2. 2Set up authentication. Complete the authentication and identity setup described in the lark-shared skill, which is not included in this package.
  3. 3Install the skill. Unzip the downloaded package into ~/.claude/skills/ or a project's .claude/skills/ folder, or upload the ZIP under Settings → Capabilities → Skills in the Claude apps.
  4. 4Make a natural request. Describe the Feishu or Lark task and state that no existing command covers it, so the agent knows to search the native API documentation.
  5. 5Review the endpoint. Check the method, path, parameters and required scopes the agent reports before approving the call.
  6. 6Confirm write or delete calls. For any POST, PUT or DELETE request, confirm your intent and, where supported, preview with --dry-run before the real call runs.

What is an agent skill, and what does this one do?

An agent skill is a folder containing a SKILL.md file that gives an AI agent instructions for a specific job. When a request matches the skill, the agent reads those instructions and follows them. This skill's SKILL.md is the only file in the package.

The skill is a fallback. First the agent checks whether an existing lark-* skill or a registered lark-cli command already covers the task. If not, it works down the Feishu/Lark documentation in layers: it fetches the top-level index at open.feishu.cn/llms.txt (or open.larksuite.com/llms.txt for Lark), then a module document, then the page for the specific API. From that page it takes the HTTP method, path, parameters, permissions and error codes, and then calls the endpoint with lark-cli api.

  • Checks existing CLI commands first (for example lark-cli im --help)
  • Finds the endpoint in the official docs instead of guessing it
  • Calls it with lark-cli api GET, POST, PUT or DELETE
  • Asks you to confirm intent before any write or delete call

Install the skill from a ZIP or the Claude apps

Download the ZIP from the directory page and unzip it. For a personal install, place the lark-openapi-explorer folder in ~/.claude/skills/. For a single project, place it in that project's .claude/skills/ folder.

If you use the Claude apps, you can upload the ZIP instead. Open Settings, go to Capabilities, then Skills, and upload the package there.

The skill declares that lark-cli must be installed and available on the command line. The source does not give an installation command for lark-cli, so follow the lark-cli documentation for that step.

  • Personal: ~/.claude/skills/lark-openapi-explorer/
  • Project: .claude/skills/lark-openapi-explorer/ inside the project folder
  • Claude apps: Settings → Capabilities → Skills, then upload the ZIP

Set up authentication before you start

The skill depends on the lark-shared skill for authentication, identity switching and global security rules. That file is not included in this package, so you need to have it set up separately. Without a working lark-cli authentication setup, calls will fail or may act on the wrong tenant.

The app you authorise also needs the Feishu/Lark permissions (scopes) for each endpoint you call. The skill tells the agent to list the required scopes, so check them before you run a call.

  • Install lark-cli and confirm it runs from the command line
  • Get the lark-shared skill and complete its authentication setup
  • Confirm your app has the scopes the endpoint needs

Trigger the skill with a natural request

You do not need a special command. Describe the workspace task in plain language, and say that you need the native API if the usual tools do not cover it. The agent should then decide whether this skill applies.

Requests that suit this skill are ones where no dedicated command exists. Requests for common tasks that existing skills already handle should go to those skills instead, because the skill says to use them first.

  • "Add these two users to the group chat oc_xxx. There is no CLI command for it, so find the native API."
  • "Set the announcement for this group chat. Look up the official OpenAPI endpoint first."
  • "Find the Feishu API that lets me change a group chat's announcement, then show me the exact call before running it."

Example: add members and set a group announcement

The source includes two worked examples. In the first, the agent checks lark-cli im --help, finds no create command for chat members, and uses POST /open-apis/im/v1/chats/:chat_id/members. The call adds members by ID and passes the member ID type as a query parameter.

In the second, the agent finds no announcement command and uses PATCH /open-apis/im/v1/chats/:chat_id/announcement. The IDs shown in the examples (oc_xxx, ou_xxx) are placeholders; replace them with real values from your workspace. Before running a write call like these, check the result the agent presents and confirm it is the change you want.

lark-cli api POST /open-apis/im/v1/chats/oc_xxx/members --data '{"id_list":["ou_xxx","ou_yyy"]}' --params '{"member_id_type":"open_id"}'

What the agent should report back

The skill sets an output format. A good result tells you the API name and purpose, the HTTP method and path, the key required and optional parameters, the scopes needed, the full lark-cli api command, and any notes such as rate limits or special constraints.

If you use English, the skill asks the agent to translate the documentation content into English. The source notes that the documentation itself is written in Chinese.

  • API name and one-line purpose
  • METHOD /open-apis/... path
  • Required and common optional parameters
  • Required scopes
  • Full lark-cli api command
  • Notes on limits and constraints

Safety notes: scripts and network access

This skill runs scripts and uses the network. It runs lark-cli commands and fetches documentation from open.feishu.cn or open.larksuite.com. Because the calls are real API requests, they act on your Feishu or Lark tenant.

The skill's rules are clear about writes. Before any POST, PUT or DELETE call, the agent must confirm your intent. It should also preview the request with --dry-run where the API supports it. The agent is told not to guess paths or parameters. For sensitive actions such as removing members or deleting groups, it should explain the scope of the effect first. Read the proposed command yourself before you approve it.

  • Write, update and delete calls require your confirmation
  • Use --dry-run to preview where supported
  • Do not approve a call whose path or parameters you did not check
  • Removing members or deleting groups can affect many people

What people use it for

  • Adding members to a group chat when lark-cli has no create command for chat members
  • Setting or updating a group chat announcement through the native announcement endpoint
  • Finding the exact method, path and scopes for a less common workspace API before writing your own script
  • Making a documented, traceable API call instead of a guessed one, so a teammate can review the same endpoint

Frequently asked questions

Is the lark-openapi-explorer skill free and what is its licence?

The skill is listed under the MIT licence in the source. The source does not state pricing for Feishu, Lark or lark-cli, so check those services' terms separately.

What do I need before I can use this skill?

You need lark-cli installed and available on the command line, the authentication setup from the lark-shared skill, a Feishu or Lark account, and app permissions (scopes) for the endpoint you call. The skill also needs network access to open.feishu.cn or open.larksuite.com.

How do I install lark-openapi-explorer?

Download the ZIP from the directory page and unzip it into ~/.claude/skills/ for personal use or into a project's .claude/skills/ folder. In the Claude apps, you can upload the ZIP under Settings → Capabilities → Skills.

Why does the agent check existing skills before searching the docs?

The skill's steps say to confirm the existing capability is missing first, using a command like lark-cli im --help. If a command or shortcut already exists, the agent should use it and not continue searching.

Does the skill work with Lark as well as Feishu?

Yes. The source lists open.feishu.cn for Feishu and open.larksuite.com for Lark. The skill says to default to Feishu if the brand is unclear.

How is this different from using lark-cli commands directly?

Regular lark-cli commands wrap specific endpoints. This skill is for endpoints with no wrapper, where it finds the native OpenAPI path in the docs and calls it through lark-cli api. It is a fallback rather than a replacement for existing commands.

New agent skills, repos and AI tools — every week

The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.

Our verdict

Use this skill if you automate Feishu or Lark workspaces and need a native endpoint that lark-cli does not wrap, and you want each call traced to the official documentation. Pick an existing lark-* skill or lark-cli command for common tasks, and avoid letting an agent run bulk or destructive operations without checking each call yourself.

Download the skill