BucketDesk

Asks a question about one document (PDF, Word, text and similar) using the connection's own AI model in the customer's AWS account. Returns the answer with citations and a chat_id; pass chat_id with a follow-up question to keep the context. Costs AI usage and needs a write token.

ask-document

Lists one page of folders and files in a bucket scope. Pass the returned next cursor to get the next page. starts_with finds names beginning with some text inside the folder.

browse

Shares a file or folder with one email address. BucketDesk emails the recipient a sign-in code; the link only works for that address and stops working when it expires. Folders can only be shared with members of the workspace. Needs a write token.

create-share-link

Gets one file by an id from search: its text for text formats (txt, md, csv, json, code and similar, up to 1 MB) and its details. For PDFs and Office files, use ask-document.

fetch

Gets an archive job's status and results. A finished compress job includes a download link that works for ten minutes.

get-archive-job

Gets one file's details: size, type, last modified, storage class, whether it must be restored first, and its etag.

get-metadata

Gets when, where (latitude and longitude) and on what camera a photo or video was taken, plus its dimensions and video length, read from the file's own EXIF or QuickTime data.

get-media-metadata

Lists the workspace's storage connections (AWS accounts) and the bucket scopes in each. Other tools take a connection id and a scope id from here.

list-connections

Searches file names in every connected bucket for all words of the query (case-insensitive). Returns ids to pass to fetch.

search

Starts an archive job: "inspect" lists what is inside a .zip, "extract" unpacks it into the bucket, "compress" zips a folder for download. Returns the job; check it with get-archive-job. Needs a write token.

start-archive-job