lark-doc

Use lark-cli docs +fetch with the document URL or token. You can read the whole document or narrow scope with outline, section, range, or keyword modes, and choose detail levels simple, with-ids, or full depending on whether you need block IDs for editing.

Follow the creation workflow: define a Presentation Decision JSON, run docs +script init-draft to get a workspace, write the XML draft, validate it with the parse command, then publish with docs +create. Simple imports of existing content can skip straight to +create.

Yes. Fetch the target area with --detail with-ids to get block IDs, then use docs +update commands like str_replace, block_replace, block_insert_after, or block_delete for minimal scoped changes. Avoid overwrite unless a full rebuild is intended, since it can drop comments and resources.

Yes. Images can be inserted from local files, the system clipboard, or public HTTPS URLs via docs +media-insert or XML img tags. Whiteboards can be created inline from Mermaid, SVG, or PlantUML sources, and existing boards are edited through the lark-whiteboard skill.

Drive-level file operations like searching, copying, permissions, and import/export belong to lark-drive, and standalone comment management also routes there. Spreadsheet and Bitable data operations are handled by lark-sheets and lark-base respectively.

Clipboard insertion on Linux requires xclip, wl-paste, or xsel to be installed. If the clipboard contains no image or the tools are missing, the command reports no image data; provide a local file path with --file instead.