# Manage Lark Docs and Feishu Wiki Content

Teams need reliable document updates without switching between tools. This skill guides Claude, Codex, and Claude Code through authenticated lark-cli workflows for Lark Docs.

## Install

```bash
npx skillstore add larksuite/lark-doc
```

## Metadata

- Status: approved
- Slug: larksuite-lark-doc
- Version: 2.0.0
- Author version: 2.0.0
- Skillstore revision: r2
- Version status: valid
- Tree hash: 0131bc55084c6812878982b2efd95d661f28e44ef17f32b02a4189dcbbc567ea
- Author: larksuite
- GitHub username: larksuite
- License: MIT
- Repository: https://github.com/larksuite/cli/tree/main/skills/lark-doc/
- Ref: ebdfe608f5de2b66ff37ab4af12af8ac4f5e8006
- Supported tools: Claude, Codex, Claude Code
- Audit status: complete
- Agent install advisory: allowed
- Manual install advisory: allowed
- Artifact signature: available
- Audit attestation: unavailable
- Human verification: not\_verified
- Risk factors: network, filesystem, external\_commands
- Quality score: 87
- Quality tier: featured
- Public page: https://skillstore.pages.dev/skills/larksuite-lark-doc
- Manifest: https://skillstore.pages.dev/api/skills/larksuite-lark-doc/manifest

## Capabilities

- Fetch Lark Docx or Wiki content by URL or token with XML, Markdown, scoped ranges, outlines, keywords, and block identifiers.
- Create documents from XML or Markdown and return the resulting document link and identifiers.
- Update text or blocks with append, overwrite, replacement, insertion, deletion, movement, and copying operations.
- Insert, preview, or download document media, download whiteboard thumbnails, and manage Docx cover images.
- List document history, select versions, start reverts, and monitor revert status.
- Read existing Mindnote nodes, create or update nodes, and calculate document word or character counts.

## Use Cases

- Summarize a Document Section: Fetch only the requested section and produce a focused summary with relevant block context.
- Maintain Existing Documentation: Locate target blocks, apply precise edits, preserve embedded resources, and verify the updated content.
- Publish an Operations Document: Create a structured document from a brief, then add approved media and validate the final length.

## Prompt Templates

### Read a Document

```
Read this Lark document and summarize its main points. Use the smallest useful fetch scope: [document URL].
```

### Create a Document

```
Create a Lark document about [topic]. Use clear headings, concise paragraphs, and the requirements in [brief].
```

### Edit a Section

```
Update only [section name] in this document. Preserve existing mentions, media, resource tokens, and unrelated blocks: [document URL].
```

### Refine and Verify a Long Document

```
Review this document for [audience]. Fetch its outline, edit sections sequentially, preserve resources, and verify the requested word count: [document URL].
```

## Limitations

- Requires lark-cli, an authenticated Lark account, and permission for each target document or resource.
- Does not manage document comments, cloud storage files, or Drive permissions directly.
- Embedded Sheets, Base tables, notes, and whiteboards require related Lark skills for internal operations.
- Mindnote commands only read existing Mindnotes or change their nodes; creating a new Mindnote uses the whiteboard workflow.

## Best Practices

- Confirm user intent before write, delete, media transfer, cover change, Mindnote update, or history revert operations.
- Fetch outlines or sections before large documents to limit data access and preserve context.
- Preserve resource tokens, mentions, embedded objects, and media unless the user explicitly requests changes.

## Anti Patterns

- Do not fetch an entire document when a section, keyword, or range fetch is sufficient.
- Do not overwrite a document when the user requested a precise section edit.
- Do not guess local file paths, clipboard contents, block identifiers, or history versions.

## Security Audit

- Audited at: 2026-07-23T16:08:34.702\+00:00
- Summary: All 181 alerts are false positives from readable Markdown, placeholders, relative documentation links, and parser syntax. No prompt injection, hidden payload, or malicious command construction was found. The skill still performs authenticated reads, writes, downloads, uploads, and reverts through lark-cli.

## Stats

- Views: 345
- Downloads: 27
- Favorites: 1
- Popularity score: 0