acknowledge_temporary_file
ChatGPTCall after a handoff file has been received and saved. The backend deletes the original immediately. Do not use it for publication or export files.
attach_document_source
ChatGPTAttaches source content (Markdown or HTML) to a registered document version, replacing any source already attached. Split pages with ## p.N Title; write chart values as json code blocks that carry page and order, and they become the page text of the AI page. First upload the file as a purpose=publication temporary file and pass its verified fileId (up to 512KB). The result's pageCheck reports format problems.
attach_mobile_summary
ChatGPTAttaches a mobile summary page (HTML) to a registered document version, replacing any summary already attached. Build the HTML as described by featpaper_guide(topic=mobile-summary), upload it as a purpose=publication, mimeType=text/html temporary file and pass the verified fileId (leave the preview kit out; the server strips it anyway). Every section needs the PC PDF page numbers (data-page). If check.ok is false, fix check.problems and attach again. numbers.mismatch is only a warning and the summary is already attached: fix only figures used in the summary that differ from the source, not section ordinals (01, 02) or other non-figures. A failed attach does not affect the registration or its links.
cancel_document_upload
ChatGPTCancels a request that has not been registered yet. It does not delete any existing document.
complete_document_upload
ChatGPTWith variant, verifies only that transferred PDF. Without it, checks all required PDFs and accepts the document registration once.
complete_temporary_file
ChatGPTAfter the file is sent, verifies its real size, type and SHA256 on the backend. Registration (publication) files are verified as account-owned files and take no accessKey; keep the accessKey only for guest export follow-up calls. Use only verified fileIds for document registration.
continue_space_analytics
ChatGPTReads the next part of a space statistics read started by get_space_analytics (jobId) and adds it to the totals without double counting.
create_data_room
ChatGPTCreates a private draft data room.
create_document_publication
ChatGPTAfter login, registers a reviewed, account-owned PDF asynchronously. A new AI viewer document takes desktop and mobile PDFs together (mobileRequired=true) or a desktop PDF alone (mobileRequired=false). With purpose=upload it registers a new document or replaces the version of an existing one (originKey); other purposes apply sharing settings. The authored source (Markdown or HTML) uploaded with the PDFs is passed as sourceFileId and attached when registration becomes ready (see sourceAttach in the result). A Markdown source the AI rebuilt from a PDF the user gave is passed with sourceOrigin=reconstructed_from_pdf: it is checked against the text of the PDF (same page count, at least 60% of each page) and attached with a provenance note only when it matches; otherwise the PDF registers alone (source.origin and sourceReconstruction in the result). The call waits briefly for ready and returns that status; when it is ready the result starts with the share link and draws the link card itself, so no status call or show_document_link is needed. Temporary files that were uploaded but not yet verified are verified here, so complete_temporary_file can be skipped for registration files. If spaceId is omitted the last used space is used. A document written or rebuilt by AI (viewerAuthoring) registers only with sourceFileId, and the source is checked before registration; a PDF alone is registered only when the user uploads a PDF they already have. A document made by AI in this conversation has no source-less option: it passes viewerAuthoring and sourceFileId, and a PDF already uploaded with its source, carrying the authoring marker, or whose page text matches a recent source (re-saved or converted copies included) is not registered without it. It first looks for an existing job with the same file, input and key and replays it. outcome_unknown means: look it up with the same key. ready is not the same as PDF conversion finished.
create_file_export
ChatGPTCreates an XLSX, CSV, DOCX or MD file from a table or Markdown without login. It is not a PDF or PPTX converter. Before an unauthenticated call, generate a 32-byte random accessKey in the execution tool and keep it; when retrying keep the same accessKey, idempotencyKey and input, and never show the key or signed URLs to the user.
create_folder
ChatGPTCreates a new folder under the specified parent folder.
create_share_link
ChatGPTCreates a document share link that identifies one recipient and returns that link (kind, label, email, url). Checks the server-side edit permission. Anyone holding the link can open the document.
delete_data_room
ChatGPTDeletes a data room.
delete_document
ChatGPTDeletes the document the user specified. Confirm the target with the user first.
delete_document_communication
ChatGPTDeletes the per-document inquiry and call-to-action settings.
delete_document_leadform
ChatGPTDeletes the document's information-entry form.
delete_folder
ChatGPTDeletes a folder. Check the effect on the items it contains first.
delete_share_link
ChatGPTDeletes a user-created share link. System links cannot be deleted.
delete_temporary_file
ChatGPTDeletes an owned temporary file that is no longer needed. Do not delete the input file of a registration job that is in progress.
duplicate_document
ChatGPTCopies a document into the specified folder. Creates a new document and leaves the original unchanged.
ensure_ai_search_link
ChatGPTReturns the public AI search link of a document (kind, label, url), creating it if it does not exist. The link is publicly accessible.
ensure_leadform_link
ChatGPTReturns the document's information-entry link (kind, label, url), creating it if it does not exist. Anyone holding the link can open the form.
featpaper_get_panel_export
ChatGPTPanel only: after the backend finishes generating the file, delivers the download address through widget metadata.
featpaper_guide
ChatGPTReads the FeatPaper guide without login. topic=viewer-authoring returns the PDF authoring and transfer guide for new viewer documents and attachment reconstruction, including the desktop and mobile PDF specification and the general registration flow. topic=mobile-summary returns the authoring instructions and preview kit for the phone-sized mobile summary page (HTML). Other topics look up the official how-to for sharing, download limits and reader analytics; for pricing it points to the official website.
featpaper_open_document_panel
ChatGPTOpens a public FeatPaper document in a side panel next to the conversation without login, so the user can page through it. Use the read tools to analyze the content. File creation and download work without login; account actions need login.
featpaper_read_attached_pdf
ChatGPTReads an attached PDF page by page as text without storing or publishing it. Only HTTPS PDFs from supported hosts are accepted; if next_from is present, keep reading from it.
featpaper_read_document
ChatGPTReads a FeatPaper document shared with the user, with page numbers, to summarize it, answer questions, translate it or extract tables and quotes. If nextFrom is present, keep reading from it. Cite the returned url. With only a document ID (documentId), the address is /n/{documentId}; /u/ is for share-link codes (/u/{code}/{suffix}). If domainUnverified is true, custom-domain ownership was not verified and the document was read as the public document with the same key in the current environment; say so.
featpaper_save_page_image
ChatGPTPanel only: displays the selected page or region as an image. The file is saved in the browser; no server temporary file is created.
featpaper_save_page_table
ChatGPTPanel only: extracts the original text as a table without login and accepts a backend XLSX generation job. Keep the accessKey you stored before the call and the same idempotencyKey.
featpaper_show_translation
ChatGPTShows the original page image next to a page-by-page translation of a document that was already read. This tool does not produce the translation.
featpaper_view_page
ChatGPTShows pages or a selected region of a FeatPaper document as images, up to 4 pages per call. Do not guess image content that could not be checked.
get_account
ChatGPTReturns the connected FeatPaper account and the last-used space.
get_data_room
ChatGPTReturns a data room's settings and the document links it contains.
get_data_room_usage
ChatGPTReturns the data room creation limit and usage.
get_document
ChatGPTReturns document details and the current account's access permission.
get_document_ai_insights
ChatGPTReturns the AI insights that have already been generated for a document.
get_document_communication
ChatGPTReturns the inquiry and call-to-action settings of a document.
get_document_download
ChatGPTIssues a short-lived download URL for a document under the current permissions. Treat the URL as private: do not repeat it to the user or write it to logs.
get_document_leadform
ChatGPTReturns the information-entry form settings of a document.
get_document_publication_status
ChatGPTChecks the registration job and the PDF conversion status. For an uncertain outcome check the reconciliation and do not register again.
get_document_share_summary
ChatGPTSummarizes the current document's share links, access conditions and inherited settings. The backend checks edit permission on the current version; no link is created.
get_document_source
ChatGPTReturns the format, size and attach time of the source content attached to a document version. It does not return the content itself.
get_document_summary
ChatGPTReturns the existing analysis summary of a document.
get_document_upload_context
ChatGPTReturns the session, space and limits for the file-selection screen.
get_document_upload_destination
ChatGPTReturns the registration folder and usage of the selected space.
get_document_upload_folder
ChatGPTReturns the sharing scope and edit permission of the selected folder.
get_document_upload_status
ChatGPTReports registration and conversion status separately. Do not re-register on outcome_unknown; check it instead.
get_file_export_status
ChatGPTChecks the status of a file export. When ready, call get_temporary_file with result.fileId to download it.
get_folder
ChatGPTReturns a folder's subfolders and documents.
get_folder_breadcrumb
ChatGPTReturns the location path of a folder.
get_folder_permission
ChatGPTReturns the current user's permission on a folder.
get_folder_tree
ChatGPTReturns the tree of accessible folders in a space.
get_mobile_summary
ChatGPTReturns the mobile summary page attached to a document version and its number-check result. Use it to see whether a summary is attached and what it contains.
get_share_link_settings
ChatGPTReturns a share link's effective settings, inheritance and access control.
get_space
ChatGPTReturns space details and feature limits.
get_space_analytics
ChatGPTReads view statistics for a whole space as totals: unique, anonymous and lead counts, traffic sources (referrer, OS, browser, country) and the most viewed documents with their view counts. It returns counts only, no per-person records. A large space is read in parts: when complete=false, continue_space_analytics reads the next part with the returned jobId.
get_space_usage
ChatGPTReturns the server-calculated plan and upload usage with limits for a space.
get_temporary_file
ChatGPTReturns the status and a short-lived download address of an account-owned file, or of a guest file whose permission is proven by its accessKey. In local_script mode only the execution tool should use the URL. Opening the download link is not a receipt acknowledgement; export and publication files are removed by explicit deletion or expiry.
list_data_rooms
ChatGPTLists the data rooms of a space.
list_document_shared_users
ChatGPTLists the users a document is shared with.
list_document_versions
ChatGPTLists the versions that belong to the document's origin group.
list_documents
ChatGPTLists the documents of a space. If documents is empty but nextCursor is present, keep paging.
list_folder_shared_users
ChatGPTLists the users a folder is shared with.
list_folders
ChatGPTLists the folders and drives of a space.
list_my_documents
ChatGPTLists the documents owned by the connected user in a space.
list_share_links
ChatGPTLists a document's default, recipient and lead-form links.
list_space_members
ChatGPTLists the members of a space.
list_spaces
ChatGPTLists all spaces the connected account can access.
move_document
ChatGPTMoves a document to the specified folder.
move_folder
ChatGPTMoves a folder under another parent folder.
prepare_browser_document_export
ChatGPTOpens a screen that converts HTML to PDF or PPTX in the browser. The HTML must contain a single deck-stage with width="1920" height="1080" and 1 to 60 direct child section[data-label] elements, with inline CSS and images. Inline HTML is limited to 512KiB and a temporary text/html fileId to 10MiB. The conversion and download happen in the user's browser when they press the button; no server-side Chromium is used. With destination the user can register the PDF on FeatPaper after conversion. Give the user the returned conversionUrl. Do not claim the conversion is complete from the tool response alone.
prepare_document_upload
ChatGPTAfter login, opens a screen where the user picks a PDF from their device and registers it, and returns a sessionId and uploadUrl. It is for a PDF the user already has that cannot be passed through an execution tool or upload_conversation_files. With picker=in_chat the card in the chat shows the space, folder and file picker and registers the file; without it the card shows progress and the file is picked on a browser page. The screen takes PDF files. A document written in this conversation is uploaded with its authored source through upload_conversation_files or prepare_temporary_file and registered with create_document_publication(sourceFileId). The requested title, purpose, recipients and viewerStyle are kept through registration; sharing purposes need the featpaper:share scope. Check progress with get_document_upload_status. Passing originKey uploads a new version of an existing document and keeps its share settings.
prepare_document_upload_transfer
ChatGPTIssues the PDF transfer target when the user presses register.
prepare_temporary_file
ChatGPTPrepares a FeatPaper temporary upload from a local file's name, real byte size and SHA256 and returns a signed multipart POST address and fields. After sending the file, verify it with complete_temporary_file (registration files may skip it: create_document_publication verifies them). purpose=publication (for registration) and handoff need OAuth login and the featpaper:manage scope and create account-owned files; only purpose=export works without login and stays valid for 24 hours. A handoff file is deleted when receipt is acknowledged and expires after 1 hour. The registration order for new documents is in featpaper_guide(topic=viewer-authoring). Use it only where your execution environment can send HTTPS requests to the returned address; ChatGPT's code sandbox has no network, so in ChatGPT use upload_conversation_files instead. This call is not idempotent: calling it again with the same file creates a new temporary file.
preview_document_publication
ChatGPTChecks the registration location and sharing settings for a verified, account-owned PDF after login. Registration uploads (publication or handoff) need OAuth from the preparation step on. A mobile PDF is linked to the same document with mobileFileId. Calling create_document_publication with the same file, input and idempotencyKey registers exactly what was reviewed.
read_result
ChatGPTReads a large result in pieces without loss. Concatenate the pieces in order and read them as JSON.
reconcile_document_upload
ChatGPTMatches an upload request whose outcome is uncertain with a document found in the document list. Links them only when the verified file IDs match; never registers again.
remove_document_source
ChatGPTDeletes the source content attached to a document version.
remove_mobile_summary
ChatGPTDeletes the mobile summary page attached to a document version.
rename_document
ChatGPTChanges a document's title, replacing the previous title.
restore_document_version
ChatGPTRestores the selected version. This changes the current version of the document.
retry_document_upload
ChatGPTRetries only the PDFs that failed before registration. With variant, other files that are already verified are kept.
rotate_leadform_link
ChatGPTReplaces the information-entry link, invalidates the previous one and returns the new link (kind, label, url).
save_data_room_links
ChatGPTReplaces the links and groups contained in a data room.
save_data_room_settings
ChatGPTChanges a data room's name, brand and contact settings, replacing the previous values.
save_document_communication
ChatGPTSets the inquiry and call-to-action contact of a document, replacing the previous settings.
save_document_leadform
ChatGPTSaves the document's information-entry form settings, replacing the previous settings.
search_documents_and_folders
ChatGPTSearches the accessible folders and documents by keyword.
set_data_room_visibility
ChatGPTPublishes a data room or switches it back to a private draft. Publishing makes it reachable through its public address.
set_document_share_settings
ChatGPTChanges a document's download, expiry, email restriction, information-entry form, viewer style, QR, inquiry and search-listing settings, replacing the previous values. For an email restriction pass isPrivate and receiver together. Search listing is inherited by public links, so change it only when the user explicitly asks.
set_share_link_access
ChatGPTSets the password and the email allow list of a share link, replacing the previous values.
set_share_link_disabled
ChatGPTDisables or re-enables the specified share link. Disabling revokes access through that link.
set_share_link_settings
ChatGPTChanges the sharing, expiry and notification settings of one link, replacing the previous values. null re-inherits the document setting.
share_document
ChatGPTGrants space users access to a document. This widens who can read the document.
share_folder
ChatGPTGrants the specified users access to a folder. This widens who can read the folder's contents.
show_document_analytics
ChatGPTReturns the title of a document and the link to its analytics page on featpaper.com, where the FeatPaper account that can access the document sees its viewing statistics after sign-in. Only the title and the link are returned; no viewer data.
show_document_link
ChatGPTShows the share link of a registered document as one card (view and copy). It does not create a link. The registration result (create_document_publication) already draws the same card, so use this only when the link is asked for again later.
show_document_sharing
ChatGPTOpens a screen to review and change a document's current share links and settings, for when the user asks to see or change sharing settings. It is not how a registration or a newly created link is shown (that is the one link just made). Opening the screen does not create a link.
unshare_document
ChatGPTRemoves user sharing of a document, revoking their access.
unshare_folder
ChatGPTRemoves folder sharing for the specified users, revoking their access.
update_document_memo
ChatGPTReplaces the management memo of a document. It does not edit the document content.
update_document_share_permission
ChatGPTChanges the permission of a user the document is shared with, replacing the previous permission.
update_folder
ChatGPTChanges a folder's name or memo, replacing the previous values.
update_folder_share_permission
ChatGPTChanges the permission of a user the folder is shared with, replacing the previous permission.
upload_conversation_files
ChatGPTUploads the files the user gave in this conversation as FeatPaper temporary files and verifies them. Used by hosts that can pass files as tool arguments (such as ChatGPT) when the execution environment cannot send them to the upload address; it replaces the prepare_temporary_file and complete_temporary_file steps. file is the PC PDF, mobile_file the mobile PDF, source_file the authored source (Markdown or HTML), mobile_summary_file the mobile summary HTML. Upload the PDF and the source (and summary) together. Pass the returned fileId, mobileFileId and sourceFileId to create_document_publication, which attaches the source when registration finishes; pass mobileSummaryFileId to attach_mobile_summary after registration. If only the source or summary is invalid, the PDF IDs are still returned with sourceError or mobileSummaryError. A document written or rebuilt by AI registers only with its source, so fix the source and upload source_file alone (file can be omitted). A source rebuilt from a PDF the user gave is uploaded with source_origin=reconstructed_from_pdf, and the server checks it against that PDF. If login is required this call reports it. Registration and sharing happen in create_document_publication.