Lark CLI

SkillDocs & knowledge

Lark/Feishu workspace operations through the `lark-cli` tool — calendar and meeting rooms, docs, sheets, Base (多维表格), wiki, drive, mail, IM messages, tasks, OKR, approvals, minutes, video conferences, whiteboards, slides, contacts, attendance, and real-time events. Use whenever the user mentions 飞书 / Lark / 我的日程 / 约会议 / 订会议室 / 飞书文档 / 多维表格 / 知识库 / 云空间 / 发消息 / 群聊 / 邮件 / 待办 / OKR / 审批 / 妙记 / 会议纪要 / 画板 / 幻灯片, or asks to look up a colleague, or hands you a `feishu.cn` / `larksuite.com` URL. Also covers `lark-cli` setup, auth login, and permission-denied errors.

Available today. Use it from your connected AI after setup.

Connect ahel once, and every AI you use reads what you have installed.

Then ask your AI: use the Lark CLI skill

What this skill tells your AI

The instructions your AI receives, as published by vaayne/agent-kit in skills/lark-cli/SKILL.md and read by ahel’s review.

lark-cli carries its own agent documentation, compiled into the binary and versioned with it. Read it from the CLI; do not work from memory and do not look for docs in this repo.

Routing

lark-cli --help                      # the 23 domains, one line each
lark-cli <domain> --help             # +shortcuts (prefer these) and raw API resources
lark-cli skills read lark-<domain>   # the domain guide: concepts, workflows, gotchas
lark-cli skills read lark-<domain> references/<file>.md   # deeper references
lark-cli schema <service>.<resource>.<method>             # params, types, scopes

Start with lark-cli --help to pick the domain, then read that domain's guide before the first call. lark-cli api <METHOD> <path> is the escape hatch when no typed command exists.

Read lark-cli skills read lark-shared first for auth, --as user vs --as bot, and permission-denied handling. Identity does not carry across commands: pass --as explicitly on every call in a workflow.

Rules

Every command's --help labels it read, write, or high-risk-write. high-risk-write requires --yes, and only after the user has confirmed. Use --dry-run to preview a request without sending it.

--jq <expr> filters JSON output. Suppress the _notice banner that otherwise pollutes every JSON response:

LARKSUITE_CLI_NO_UPDATE_NOTIFIER=1 LARKSUITE_CLI_NO_SKILLS_NOTIFIER=1 lark-cli ...

Signals

GitHub stars
53
Forks
4
Last commit
Sep 2026
Advanced
Catalog kind
skill
Gateway key
lark-cli-vaayne
Source
github.com/vaayne/agent-kit