ソース情報
- リポジトリ
- tubecreate/tubecli
- ソースの最終更新活動
- 2026年9月2日 06:05
- 検出された SKILL.md の言語
- 英語
- スター
- 167
- フォーク
- 3
インストール方法
デフォルトでは、最初にソースを確認する Prompt が選択されています。直接コマンドに切り替えるか、ローカルコピーをダウンロードすることもできます。
ソースファイルを確認
インストールを決める前に、SKILL.md と SkillsMP に表示されている付属ファイルをお読みください。
メニュー
デフォルトでは、最初にソースを確認する Prompt が選択されています。直接コマンドに切り替えるか、ローカルコピーをダウンロードすることもできます。
インストールを決める前に、SKILL.md と SkillsMP に表示されている付属ファイルをお読みください。
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
直接コマンドでは確認用 Prompt が省略されます。実行前にソースを確認してください。
npx skills add https://github.com/tubecreate/tubecli --skill file-managerコマンドは1行のまま表示されます。コピー前に横へスクロールして全体を確認してください。
ローカルで確認しますか?SkillsMP が現在取得できるファイルをダウンロードできます。
Manage anti-detect browser profiles, open/close browsers, automate the web
Manage OAuth credentials & tokens for Google, Facebook, TikTok
Manage and deploy Cloudflare Workers websites. Provides a card list of websites, manual add, and automatic creation from templates (git clone → build OpenNext → wrangler deploy → D1 + R2). Integrates with cloud_api for Cloudflare credentials (CF_API_TOKEN + CF_ACCOUNT_ID, or Global API Key + email).
SKILL.md を表示中
| name | File Manager |
| description | Manage files and folders on the computer |
This extension lets the AI create, delete, move, copy and list files/folders directly on the user's computer.
{"action": "file_action", "operation": "create_folder", "path": "~/Desktop/my_folder"}
{"action": "file_action", "operation": "create_file", "path": "~/Desktop/note.txt", "content": "File content"}
{"action": "file_action", "operation": "list", "path": "~/Desktop"}
{"action": "file_action", "operation": "delete", "path": "~/Desktop/my_folder"}
{"action": "file_action", "operation": "move", "path": "~/Desktop/old_name.txt", "destination": "~/Desktop/new_name.txt"}
{"action": "file_action", "operation": "copy", "path": "~/Desktop/file.txt", "destination": "~/Documents/file_copy.txt"}
{"action": "file_action", "operation": "read", "path": "~/Desktop/note.txt"}
The extension can also manage the user's Google Drive through Auth Manager
tokens. Pass cred_id when the agent's system prompt names one — that pins the
operation to the Google account the user selected. Without cred_id the newest
authorized Drive token is used.
IMPORTANT: Google Drive files are CLOUD resources — NEVER use file_action
for them. Use the drive_* actions below. file_id values come from a
previous drive_list result.
{"action": "drive_list", "folder_id": "root", "cred_id": "<from agent auth guide>"}
Search by name: {"action": "drive_list", "query": "report", "cred_id": "..."}
{"action": "drive_upload", "local_path": "~/Downloads/video.mp4", "folder_id": "root", "cred_id": "..."}
{"action": "drive_download", "file_id": "<drive file id>", "dest_dir": "~/Downloads", "cred_id": "..."}
{"action": "drive_rename", "file_id": "<drive file id>", "new_name": "new-name.mp4", "cred_id": "..."}
Anyone with the link: {"action": "drive_share", "file_id": "...", "role": "reader", "cred_id": "..."}
A specific person: {"action": "drive_share", "file_id": "...", "type": "user", "email": "person@gmail.com", "role": "writer", "cred_id": "..."}
{"action": "drive_mkdir", "name": "Backups", "parent_id": "root", "cred_id": "..."}
{"action": "drive_delete", "file_id": "<drive file id>", "cred_id": "..."}
These edit .xlsx / .xlsm workbooks IN PLACE (other sheets, formatting and
formulas are kept) and read/append .csv. Use them for spreadsheets instead of
file_action read/create_file, which flattens the file to text.
{"action": "xlsx_read", "path": "~/Downloads/plan.xlsx", "sheet": "Sheet1", "max_rows": 100}
sheet is optional (active sheet). The reply shows | a | b | rows, the total
row count and the other sheet names. Formula cells show their last computed
result; the server does not calculate formulas, and after xlsx_append /
xlsx_write the cached results are gone, so those cells show the formula text
(e.g. =SUM(B2:B9)) and the reply says how many. Compute such totals yourself
from the rows when you need the number.
{"action": "xlsx_append", "path": "~/Downloads/plan.xlsx", "sheet": "Sheet1", "rows": [["a", "b"], ["c", "d"]]}
{"action": "xlsx_write", "path": "~/Downloads/plan.xlsx", "sheet": "Sheet1", "cells": {"A1": "v", "B2": 3}}
A block works too: {"action": "xlsx_write", "path": "...", "rows": [[1, 2], [3, 4]], "start": "B2"}.
CSV files support xlsx_read and xlsx_append only.
Boundary: when the agent belongs to a group (a GROUP WORKSPACE block in the
prompt), ONLY the spreadsheet files and folders listed there can be used, with
the access of their node (read < append < write) — nothing else exists,
even inside the sandbox below. Without a group, the sandbox below applies.
AI-triggered LOCAL file operations are sandboxed to:
~/Desktop — Main screen~/Documents — Documents~/Downloads — Downloadsdata/ — TubeCLI data folderThe human user browsing the File Manager UI is NOT limited to these folders —
only AI actions are. If the user asks about a file outside the sandbox, tell
them to use the File Manager UI, or to move it into one of the folders above.
Google Drive actions are not path-sandboxed (they act on the user's own Drive),
except drive_download's dest_dir, which must be inside the sandbox.
Use the file_manager node in the workflow builder:
command (text command), path, content, destinationresult (operation result), status