feishu-readonly-doc-export

name	feishu-readonly-doc-export
description	Export Feishu/Lark cloud documents as local files without editing remote content. Use when the user wants read-only export, backup, PDF/Word/Markdown/XLSX/CSV/PPTX/.base output, or needs to download a doc/docx/sheet/bitable/slides resource from a Feishu, Lark, Wiki, or Drive URL.

Feishu read-only document export

Use this workflow for exports only. Do not edit, move, delete, publish, or change permissions unless the user separately asks and confirms.

Required runtime reads

Before running export commands:

lark-cli skills read lark-shared
lark-cli skills read lark-drive
lark-cli skills read lark-drive references/lark-drive-export.md

If the input is a bare token, its type is unclear, or the user needs explicit resolution evidence, also read:

lark-cli skills read lark-drive references/lark-drive-inspect.md

Workflow

1. Confirm lark-cli is available and current enough:

   command -v lark-cli
   lark-cli --version
   

2. Choose the input form without manually parsing URLs:

   • Prefer --url when the user provides a URL. The current shortcut infers the type and token; Wiki URLs are unwrapped automatically.
   • For a bare token, pass --doc-type explicitly.
   • For a bare Wiki node token, use --doc-type wiki so the CLI unwraps the underlying object.
   • Use +inspect only when the type is unclear, the token source is questionable, or the final evidence needs canonical type/title/token:

   lark-cli drive +inspect --url '<feishu-url>' --format json
   

   Do not manually split a Wiki URL or pass a Wiki node token as if it were the underlying docx, sheet, or bitable token.

3. Select the format:

   Source type	Supported outputs
   docx	pdf, docx, markdown
   doc	pdf, docx
   sheet	xlsx, csv with --sub-id
   bitable	xlsx, csv with --sub-id, base; add --only-schema for a schema-only .base snapshot
   slides	pptx, pdf

4. Choose identity and paths deliberately:

   • Use --as user when exporting resources visible through the user's own Drive or Wiki permissions.
   • Use --as bot only for bot-owned resources or resources explicitly shared with the app/bot.
   • Use relative output paths under the current working directory. lark-cli rejects unsafe absolute file paths for file operations.

5. Export a URL directly with a bounded local output directory:

   lark-cli drive +export \
     --url '<FEISHU_OR_LARK_URL>' \
     --file-extension '<pdf|docx|markdown|xlsx|csv|base|pptx>' \
     --output-dir './exports' \
     --format json
   

   For a bare token, use the explicit fallback:

   lark-cli drive +export \
     --token '<TOKEN>' \
     --doc-type '<doc|docx|sheet|bitable|slides|wiki>' \
     --file-extension '<pdf|docx|markdown|xlsx|csv|base|pptx>' \
     --output-dir './exports' \
     --format json
   

   For a bitable schema-only snapshot:

   lark-cli drive +export \
     --url '<BITABLE_URL>' \
     --file-extension base \
     --only-schema \
     --output-dir './exports' \
     --format json
   

6. If the shortcut times out while the server-side task is still running, this is not failure. Continue with the returned ticket:

   lark-cli drive +task_result \
     --scenario export \
     --ticket '<TICKET>' \
     --file-token '<SOURCE_TOKEN>'
   

   Then download the returned export file_token:

   lark-cli drive +export-download \
     --file-token '<EXPORTED_FILE_TOKEN>' \
     --output-dir './exports'
   

Evidence to retain

For every export, keep these fields in the final summary:

• source URL or token type, redacted if private
• resolved doc-type
• output extension
• whether a bitable .base export included data or used --only-schema
• local output path
• whether export completed immediately or required ticket continuation
• any permission or scope errors and the exact missing scope when available

Refuse or pause

Pause and ask before proceeding when:

• the user asks to change permissions, move/delete a source file, or publish a link as part of export
• the target is private and the final artifact would be public or committed
• the only available path requires a new broad application scope instead of a narrow read/export permission

Constraints from Feishu OpenAPI

• Export is asynchronous: create export task, query task result, then download.
• Exported files are temporary; download promptly after the task succeeds.
• Enterprise watermark behavior depends on identity: app identity watermarks with app ID; user identity watermarks with user identity.
• Cloud-document access requires both API/application permission and document permission for the current bot or user identity.