Guide · Agent skill
How to use the Lark OKR Manager agent skill (lark-okr)
Updated · written with AI from the project's own documentation and checked against our catalogue data
The lark-okr skill teaches an AI agent how to work with OKRs (Objectives and Key Results, a goal-tracking method) stored in Feishu/Lark. This guide is for people who keep OKRs in Feishu/Lark and want an agent to look them up, draft them or update them. By the end you will know what the skill does, how to install it, how to phrase requests so it triggers, and what it needs to run.
Lark OKR Managerlarksuite/cli · MIT Download the skill Quick start
- 1Download the ZIP. Download the lark-okr ZIP from this page and unzip it so that SKILL.md sits inside a lark-okr folder.
- 2Place the folder. Move the folder to ~/.claude/skills/ for personal use, or to .claude/skills/ in one project.
- 3Install lark-cli. Make sure the lark-cli command is installed and runs from your terminal.
- 4Authorise your Lark account. Authorise lark-cli with your Feishu/Lark account so commands can run with --as user.
- 5Ask for OKRs in plain language. Request something like "show my OKRs for the current cycle" and let the agent run the matching commands.
- 6Review before writing. For changes, check the proposed commands and use --dry-run on batch creation where available.
What an agent skill is and what lark-okr makes the agent do
An agent skill is a folder of instructions and reference files that an AI agent reads when a task matches the skill's description. The lark-okr skill is one of these. It covers reading and editing OKR cycles, objectives (Objectives), key results (KRs), alignment relationships, quantitative indicators and progress records.
Once installed, the agent can look up a user's OKR cycles and objectives, create batches of objectives and key results, update numeric progress, write progress notes, check alignment between goals, and read or reply to OKR comments. It does this by running the lark-cli command-line tool under the user's Lark account.
- View your own or a colleague's OKR cycles and their objectives
- Create one objective or key result, or batch-create a set of them
- Update numeric progress (percentages or plain numbers) and write text progress notes
- Read, list, create, reply to and resolve OKR comments
- Check which objectives are aligned to which
Install the lark-okr skill from the ZIP
Download the ZIP for the skill from this page and unzip it. The skill folder must contain the SKILL.md file at its top level, along with its references folder.
For a personal install, place the unzipped folder in ~/.claude/skills/. To limit it to one project, place it in that project's .claude/skills/ folder instead. In the Claude apps you can also upload the ZIP under Settings → Capabilities → Skills.
- Personal: ~/.claude/skills/lark-okr/
- Single project: .claude/skills/lark-okr/ inside the project folder
- Claude apps: Settings → Capabilities → Skills, then upload the ZIP
Prepare lark-cli and your Lark account
The skill calls the lark-cli command-line tool, so lark-cli must be installed and available as a command named lark-cli. The skill's metadata lists it as its required binary. Before the agent can read or write OKRs, your Feishu/Lark account must be authorised through lark-cli. The skill runs most commands with --as user, which acts as you.
The skill tells the agent to read the shared authentication guide at ../lark-shared/SKILL.md before starting. That guide covers authorisation and permission handling. Viewing other people's OKRs may also need bot permissions, which the skill describes as optional and depends on your organisation's setup.
- lark-cli installed and on your PATH
- Feishu/Lark account authorised through lark-cli
- Permission to view colleagues' OKRs, if you need that (use --as bot where the skill says it applies)
lark-cli okr --helpTrigger the skill with natural requests
You do not need to name the skill. The skill is meant for requests to view or create OKRs, manage objectives and key results, or view alignment relationships. Write the request the way you would ask a colleague, and mention the cycle or objective if you have it.
The skill separates digital progress from text progress. If you say a number such as 75%, it updates a quantitative indicator. If you describe progress in words, it writes a progress record. Scores are only changed when you explicitly ask about scores, grades or ratings.
- "Show my OKRs for the current cycle."
- "Draft three objectives with two key results each for my team's Q goals."
- "Update the 'page load' key result to 75%."
- "Which objectives are aligned to my manager's objective?"
- "Reply to the open comments on my key results."
Examples of common operations
For a full view, the agent looks up the user ID, lists cycles, then loads the detail of one cycle. Batch creation takes a JSON list of objectives, each with optional notes and a list of key results. You can add --dry-run to preview the API calls before anything is written.
The skill's own examples show the underlying commands. They are copied here as they appear in the source.
- Look up a colleague by name or email with the contact search, then list their cycles
- Create objectives from a JSON file with --input @file.json
- Preview a batch with --dry-run before running it
- Update a key result's numeric progress with +indicator-update
lark-cli okr +batch-create --cycle-id 7000000000000000001 --input '[{"text":"提升产品用户体验","notes":"关注核心流程和用户反馈","krs":[{"text":"核心流程满意度达到 4.8 分"}]}]' --as userSafety notes: scripts and network access
This skill runs scripts and uses the network. Each command sends requests to Feishu/Lark using the account you authorised, so the agent can read and change your OKR data, and comments and progress records it writes are visible to others in the organisation. Delete operations, such as removing a progress record, are not recoverable according to the skill's command list.
Review the commands the agent proposes, use --dry-run for batch changes, and check the shared authentication guide before granting permissions.
- The skill reads and writes live Feishu/Lark data under your account
- Progress record deletion (+progress-delete) cannot be undone
- Use --dry-run on batch creation to preview the calls
Tips for better results
Give the agent the cycle or objective ID when you have it, because it avoids a search step. If you are unsure, ask the agent to list your cycles first. Keep percentages and text notes separate in your requests so the right record type is updated.
Remember what the skill does not cover. Task management belongs to lark-task, calendar and meeting scheduling belong to lark-calendar, and performance evaluation is out of scope.
- Start with your cycle list, then ask for detail
- Say "progress" or "completion" for numbers, and say "score" only when you mean scores
- Split very large lists into batches
What people use it for
- Drafting a set of objectives and key results for a new cycle from a list of goals
- Checking whether a team's objectives support the objectives of its manager
- Updating numeric key result progress after a weekly check-in
- Writing progress notes on objectives that have no numeric measure
- Reading and replying to OKR comments across a cycle
Frequently asked questions
How do I install the lark-okr skill?
Download the ZIP from this page and unzip it. Place the folder in ~/.claude/skills/ for personal use or in a project's .claude/skills/ folder, or upload the ZIP in the Claude apps under Settings → Capabilities → Skills.
What does the lark-okr skill require?
It requires the lark-cli command-line tool and a Feishu/Lark account authorised through lark-cli. Viewing other people's OKRs may also need bot permissions. The skill points to a shared authentication guide for these steps.
Is the lark-okr skill free?
The source describes the skill as MIT licensed. This page does not state pricing for Feishu/Lark or lark-cli, so check those services for their own terms.
Can it manage tasks or calendar events too?
No. The skill hands task management to lark-task and calendar or meeting scheduling to lark-calendar. It also does not cover performance evaluation.
Does it work outside Feishu or Lark?
No. Every command targets Feishu/Lark, so organisations that do not use it cannot use this skill.
Why did the agent change my progress instead of my score?
The skill treats progress and completion as numeric indicators or progress records, and only changes scores when you explicitly ask about scores, grades or ratings.
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 lark-okr if your OKRs live in Feishu/Lark and you want an agent to view, draft, batch-create and update them through your own account. Pick something else if you need task scheduling, calendar work, performance reviews, or if your organisation does not use Feishu/Lark.
Download the skill