lark-doc

Run lark-cli docs +fetch with the docx or wiki URL as the --doc argument. Use --scope with outline, section, range, or keyword to read only the part you need instead of the full document.

First fetch the document with --detail with-ids to get block IDs, then run docs +update with commands like block_replace or block_insert_after targeting that block ID. For simple text changes, str_replace with --pattern works without block IDs.

XML is the default and preferred format, especially for precise block-level edits, since it preserves block structure and styles. Use Markdown only when the user explicitly requests it or provides a .md file for import.

No. When fetched content contains sheet or bitable tags, the skill extracts their tokens and routes to the lark-sheets or lark-base skills for data operations. Document comments and drive file operations are also handled by lark-drive.

Fetch the document with --detail full in XML format and pipe the JSON output to the bundled doc_word_stat.py script. It reports word_count and char_count using a consistent counting standard for Chinese and English text.

Block IDs become invalid after overwrite, block_replace, or block_delete operations. Re-fetch the document with --detail with-ids to obtain fresh block IDs before continuing block-level edits.