Find and organize files in Google Drive, upload local files, download stored files, and manage sharing. Rename, move, copy, trash, or restore items while keeping track of their names and links.
Work with the user's Google Drive: files, folders, uploads, downloads, and sharing.
Review the instructions the agent will receive before you install.
Everything runs through hatch_gws_cli drive .... Raw API calls are space-separated (<resource> <method>) and take a --params JSON object for query and path values, plus a --json body when creating or editing. Run hatch_gws_cli drive <command> --help for a command's flags, and hatch_gws_cli schema drive.files.list (and the like) for a raw method's --params and --json shape. Check the shape before an unfamiliar method.
Drive needs a one-time connect before commands return data. Run hatch_gws_cli drive status. If it comes back not connected with a connect_url, post that exact URL as [Connect Google Drive](<connect_url>), then stop and wait for the user to tap it. If no URL is returned, report that connection is unavailable and stop; never invent one. Connecting is the user's step, so never open the sign-in, drive a browser to it, point the user to Settings or a Google account page, or ask for credentials. If status comes back unavailable, say Google Drive is not available on this device and stop.
Disconnect with hatch_gws_cli drive disconnect. Post [Disconnect Google Drive](<disconnect_url>) only when the command returns that URL. If status remains connected after the link was posted, re-post the same link and let the user finish in the browser; never send them to a Google account, security, or third-party-access page, and never loop on status. If no link is returned, rerun status once and report its state without inventing a link. If a later command reports an auth error, rerun status: post its connect link and wait when disconnected, retry the command once when connected, or report that Drive is unavailable when there is no usable link. Auth flows only through status and disconnect. Never hand-author credential files or run raw gws auth.
Most people have one Google account, and that is the default: run commands with no account flag. If the user names one of several linked Drives ("my work Drive"), list them with hatch_gws_cli drive accounts, match the user's words to exactly one display_name, and pass its account_id as --account <account_id>. If there is no exact match, ask which account rather than guessing or silently falling back to the default.
drive files list --params '{"pageSize":20}'. To browse inside one folder, query by its id: drive files list --params "{\"q\":\"'<folder_id>' in parents and trashed=false\",\"pageSize\":50}".drive files list --params "{\"q\":\"name contains 'budget' and trashed=false\",\"pageSize\":20}". Drive query operands use single quotes; use fullText contains 'text' to search inside content.drive files get --params '{"fileId":"<id>","fields":"id,name,mimeType,parents,webViewLink,modifiedTime,owners"}'.drive +upload ... (run drive +upload --help for its flags). Confirm the local path exists first, and drop it into a folder by passing that folder's id. +upload always creates a new file. Every upload path must be absolute and inside the home directory. Relative and /tmp paths fail.drive files update --params '{"fileId":"<id>"}' --upload <absolute_path>. Uploading an Office file into a Google-native Doc or Slides file converts it in place. The Docs and Slides skills own when to do that. Never upload into a Google Sheet. Google replaces the full contents, so the upload discards the user's other tabs. The Sheets skill formats through the Sheets API instead.drive files get --params '{"fileId":"<id>","alt":"media"}' --output <path>. The output path is the user's choice, so ask or reuse one they named.drive files create --params '{"ignoreDefaultVisibility":true}' --json '{"name":"Q3 Docs","mimeType":"application/vnd.google-apps.folder","parents":["<parent_folder_id>"]}'. Use "root" as the parent for the top level. This opts out of domain-wide default visibility; the folder still inherits its parent's sharing.drive files update --params '{"fileId":"<id>"}' --json '{"name":"Q3 Budget"}'.drive files update --params '{"fileId":"<id>","addParents":"<dest_folder_id>","removeParents":"<current_folder_id>"}'. Read the current parent from a files get first so you remove the right one. Moves use sharing approval because the destination can grant access.drive files copy --params '{"fileId":"<id>","ignoreDefaultVisibility":true}' --json '{"name":"Copy of Q3 Budget","parents":["root"]}'. Use the requested destination folder instead when specified. Omitting the destination can inherit the source's parent and uses shared-create approval.drive permissions create --params '{"fileId":"<id>"}' --json '{"type":"user","role":"reader","emailAddress":"alex@example.com"}'. Use writer only when the user asks for edit access.drive files update --params '{"fileId":"<id>"}' --json '{"trashed":true}'. Restore with {"trashed":false}.drive files delete --params '{"fileId":"<id>"}'. Only use it when the user clearly wants the item gone for good. Say plainly that it cannot be recovered; for a folder, also say that permanently deleting it removes the user's owned files and folders inside it.+upload calls. My Drive placement alone does not rule out domain-wide default visibility. Confirm before creating in a shared folder or modifying an already shared item because other people can see the changes.webViewLink returned by Drive; fetch the field when needed and never construct a Drive URL. Keep ids only in your working context to chain the next command.drive files export or otherwise pull a Google-native file's contents through Drive, even as a fallback when another skill's connection fails; say that opening its contents needs the Docs, Sheets, or Slides skill and stop. Drive handles the file itself: finding it, its details, and moving, sharing, or removing it. The binary download recipe above (alt=media) only fetches stored binary files, never Google-native ones.hatch_gws_cli schema drive.<resource>.<method>; treat any write as shared and confirm before it, because there is no seeded, tested private-write recipe for those surfaces.No reviews yet. Tell others what you thought.
No login needed. Share how it worked for you, ask a question, or note where it did not run.
No comments yet. Be the first to write one.
Create a 3D object with three.js, inspect it from every angle in an interactive viewer, and download it as OBJ + MTL or GLB for use in other 3D applications.
Maintain a private task list with deadlines, next steps, waiting parties, and blockers. Update progress and dependencies, merge duplicate items, and archive closed tasks so ongoing commitments stay clear and easy to follow.
Coordinate a task from a verifiable goal through completion. Handle work directly or split independent tasks across separate agent sessions, track progress, verify results, and arrange pull request merging when included in the goal.
Create animated HTML pages with smooth scene transitions and playback controls. An editable timeline lets users adjust scene lengths and speeds, with fixed landscape or portrait layouts prepared for video export.
Review synced Apple Health records, including steps, heart rate, sleep, and workouts. Retrieve daily summaries or detailed samples for a chosen period, and check sync status and missing data before interpreting the results.
Create and edit Word documents and templates, reorganize text, replace images, and work with tracked changes or comments. Apply proper headings and lists, then render every page to check its appearance before delivery.