dataroomHQ

Create a role in a dataroom. Every permission starts OFF. Pass only the permissions to enable, e.g. {"doc_view": true, "doc_download": true}. A role by itself grants nothing to anyone until it is attached to a share link or a member (assign_role). `doc_view` is blanket, not per-document: enabling it grants every document in the dataroom, and only a per-document override with can_view=False can withhold one. `hidden` decides only whether the role appears in the dataroom's role picker. Requires the 'role_manage' role permission.

Create a share link for a dataroom. Returns the link INCLUDING its token, and `url`: the link to give the user, exactly as returned - never build one from the token yourself. Anyone holding the token can reach the data room without authenticating, so treat it, and this tool's output, as a credential. This is the only way to get a usable link out of this tool, so capture it immediately. - email: restricts the link to one recipient, and is mailed to them unless notify=False - role_id: the role to grant (from list_roles); omitted, the default restricted-access role - name: a display name for the link - expires_in_days: default a year Needs: share_link_manage.

Delete a category and every subcategory under it. Documents are NOT deleted — they lose their category and become uncategorised; `documents_unfiled` in the reply counts them. A default category (one with a non-null `type`, seeded into every dataroom) cannot be deleted, and neither can a category with one underneath it. Requires the 'category_delete' role permission.

Delete a data room and everything in it — documents, folders, reports, members and share links. Irreversible. Confirm with the user first, naming the room. Requires the 'dataroom_manage' role permission.

Delete a document by id - irreversible. Its linked library file is deleted too, best-effort. Needs: doc_delete.

Delete a folder, every subfolder under it, and every document in them. This cannot be undone. `documents_deleted` in the reply counts every document the cascade removed. Requires the 'doc_delete' role permission.

Delete a generated report by kind and id (from list_generated_reports) - irreversible. A document already copied from it stays. The company profile cannot be deleted. Needs: each kind's own delete permission, or category_edit for a category summary.

Stop a notification rule's emails (from list_notification_rules). Requires the 'dataroom_manage' role permission.

Delete a role from a dataroom. Refused, with the counts, while members or share links still hold the role unless `force` is true. Forcing does NOT reassign them — it DELETES those memberships and share links, so those people lose access to the dataroom and circulating links stop working. The refusal names the counts, so try without `force` first. Built-in roles cannot be deleted. Requires the 'role_manage' role permission.

Edit a pitch deck (from list_generated_reports) from an instruction, e.g. "shorten slide 3" or "add a slide on pricing". Returns at once; the slides are rewritten in the background, so check list_generated_reports for its status. A deck still being generated or edited is refused. Requires the pitch-deck subscription feature and the 'presentation_update' role permission.

Mark documents uploaded once the upload widget has sent their files. Called by the upload widget; not intended for the model. At most 100 documents per call; each is finished on its own, and the reply lists `uploaded` and `failed`.

Generate a business report (metrics spreadsheet) for a dataroom. Builds the ARR schedule, cohorts, retention and customer-concentration sheets from the room's own categorised financial documents - it takes no file arguments. Runs in the background: created at once in QUEUED state, its sheets fill in over the next few minutes. Needs: business-report plan feature, business_report_generate.

Generate, or regenerate, one category's AI summary (from list_generated_reports). Returns at once and runs in the background - check list_generated_reports for its status. For all three company profile sections at once, use generate_company_profile instead. - website_url: product, team and competitors only; defaults to the room's - notes: product, team and competitors only; optional guidance - competitor_urls: competitors only - strict_competitors_only: competitors only; skip independent discovery Needs: category-summary plan feature, dataroom membership.

Generate a dataroom's company profile: Product, Team and Competitors. All three are researched from the company's website and queued as one kickoff, filling in over the next few minutes. REPLACES any existing profile for the room. - website_url: omit to use the room's own; refused if neither exists - competitor_urls: optional, researched by the Competitors section - strict_competitors_only: research only those, skipping discovery - notes: optional guidance, passed to all three sections Returns `{"website_url", "sections"}`; `sections` maps each generated section to its queued status, and is empty when the room has none of the three profile categories configured. Needs: company-profile plan feature, dataroom membership.

Generate a data deck - a metrics board deck - for a dataroom. Built entirely from the room's own ARR schedule and expense documents: no prompt, theme or slide count, since the content is the numbers rather than a written narrative. Slower than the other generators and not finished when this returns - it is accepted here and rendered afterwards. `name` defaults to "Board Deck - <dataroom name>". Needs: data-deck plan feature, board_deck_generate.

Generate an investment memo for a dataroom. Runs in the background: created at once in QUEUED state, its sections fill in over the next few minutes. `name` titles it; `notes` is free-text guidance for the generator. Needs: investment-memo plan feature, memo_generate.

Generate a pitch deck for a dataroom. Pitch decks only. Runs in the background: the deck is created at once and its slides fill in over the next few minutes. - name: defaults to "Presentation" - prompt: extra guidance for the deck's content - theme: one of Ocean Depths, Midnight Executive, Forest Canopy, Midnight Galaxy, Modern Minimalist, Golden Hour, Tech Innovation, Desert Rose, Arctic Frost. An unknown name is rejected; omit for the default. - slides: requested count, ~8-12 when omitted Needs: pitch-deck plan feature, 'presentation_generate'.

Get engagement analytics for a dataroom or one of its documents. - summary: room totals (views, unique views, downloads, time spent), then the same per document and per viewer, with each viewer's last activity. Pass `document_id` to narrow it to one document. - trend: one document's figures bucketed over time, by `granularity`, which the other breakdowns ignore. - pages: one document's per-page views, reading time and completion rate. - video: one video's average watch time and percentage viewed, plus a coverage curve over auto-sized buckets. Every breakdown but `summary` needs a `document_id`. Dates are ISO 8601. Engagement only. For what was done to the room - uploads, membership changes, deletions - use list_audit_logs. Needs: analytics.

Get details of a specific dataroom by ID. Returns id, name, description, created_at, updated_at, member_count, document_count and onboarding_status: null when the room never onboarded, IN_PROGRESS or DOCUMENTS_READY while it is still producing its own memo, business report and board deck, then COMPLETED or FAILED.

Get one document's metadata. Returns id, name, description, file_extension, size_bytes, category, folder, created_at, updated_at, and permissions (can_view, can_download, can_edit, can_delete).

Get a temporary URL to download a document's file. Valid 5 minutes, for a client that can make HTTP requests itself: the original, or a watermarked copy when the reader's role requires one. variant="watermarked" asks for that copy whatever your own role; a type that cannot be watermarked (video, audio, zip, spreadsheet) is refused. Also returns the name, extension and size. To read what a document says, use read_document instead.

Get a company's financial metrics, computed from its own documents in the room. - subscription: ARR, MRR, growth, churn, margins and related figures - cohorts: retention cohort grids, the cohort catalog, customer concentration - pipeline: sales opportunities Not a document reader: these are figures already derived from the room's financial documents, so it takes no file arguments. Needs dataroom membership and access to the documents behind the figures - the ARR schedule and expenses, or the opportunities document for `pipeline`.

Read one role's full permissions and document overrides by id (from list_roles). System roles never have overrides — the `overrides` list is empty for them.

Where an upload to this dataroom and folder would land, and whether you may upload there. Called by the upload widget each time the reader picks a dataroom or folder; not intended for the model. Same checks as show_upload_widget: a folder you can't see is not found.

Invite someone to a dataroom by email, with a role from list_roles. Needs: member_manage.

List a dataroom's audit log: who did what, when and to which resource, each entry with a human-readable description. Actions, not engagement. For who viewed what and for how long, use get_analytics. Filters: event_type ("document_created", "dataroom_member_added"), resource_type ("document", "dataroom", "memo", "role"), resource_id, source - "app" (the web app), "mcp" (an assistant over MCP), "chat" (the in-app assistant) or "system" - and start_date/end_date as ISO 8601. `limit` defaults to 50, at most 100; `offset` pages. Needs: audit_logs.

List a dataroom's document categories, sorted by path. Categories classify documents by subject; folders (list_folders) are the separate filing tree. This is the only way to see a category holding no documents. Each row: category_id (for list_documents or update_document), name, parent_category_id (null at the top level), path (ancestor names joined with "/", e.g. "Legal/Contracts"), type (non-null on the defaults every room is seeded with, which cannot be deleted), and document_count - documents you can see filed directly under it.

List all members of a dataroom. Returns an array with fields: - user_id: member's user ID - role_name: assigned role (e.g. Admin, Viewer) - joined_at: when the member was added

List all datarooms you have access to. Returns an array of datarooms with fields: - id: unique dataroom identifier - name: dataroom display name - description: optional description - created_at: when the dataroom was created - member_count: number of users with access - document_count: total number of documents - logo_url: signed link to the dataroom's logo, or null - can_upload: whether your role may upload documents to it

List who a specific document is shared with (email, permissions, status). Tokens are never included. Requires the 'share_link_manage' role permission, and document_id must be visible to your own role in this dataroom.

List every version of a document, newest first: the current one and each archived one (`is_current` tells them apart). Use an archived entry's `id` with restore_document_version. Requires the DOCUMENT_VERSION_CONTROL subscription feature and 'doc_view'.

List a dataroom's documents, optionally narrowed by name, category or folder. `query` matches part of a name, case-insensitively. It cannot be combined with the category or folder filters - a name search covers the whole room. Shows only documents you may view. Each row: id (for get_document and get_document_download_url), name, description, file_extension, size_bytes, category, folder, created_at, updated_at, and permissions (can_view, can_download, can_edit, can_delete). This returns the list to you. To show the room to the user, use show_dataroom_widget instead.

List the folder tree of a dataroom, sorted by path. Returns the tree to you. To show the room to the user, use show_dataroom_widget. Only shows folders you have permission to see. Returns an array with fields: - folder_id: folder identifier (use with list_documents or update_document) - name: folder name - parent_folder_id: the folder it sits in, or null at the top level - path: ancestor names joined with "/", e.g. "Legal/Contracts" - document_count: documents you can see directly in it, excluding subfolders

List a dataroom's generated reports, of one kind or of every kind. - memo: investment memos - presentation: model-written pitch decks - business_report: metrics spreadsheets - data_deck: metrics board decks built from the room's financials; ids are numbers - category_summary: AI summaries of the categories that can have one (cash flow, income statement, product, team, competitors), each with its status and whether a summary exists; the id is the category's The company profile is one per room and is not listed - read it with read_generated_report(kind="company_profile"). Returns `{kind: [items]}`, each with `id`, name and generation status. Without `kind`, every kind is listed and one you cannot see reads "no access". Needs: each kind's own view permission, and the plan feature for memos, pitch decks, business reports and data decks.

List everyone with access to a dataroom: members, dataroom shares, and document shares (grouped by email). Share-link tokens are never returned, at any nesting depth. Requires dataroom membership.

List a dataroom's email notification rules: which document event (opened, downloaded) emails the room's admins, and how often at most. Requires the 'dataroom_manage' role permission.

List roles available in a dataroom (id, name, is_system, permissions). Use the returned `id` with get_role and assign_role.

List share links for a dataroom (id, name, email, role, status, expiry). Tokens are never included. Use the returned `id` with revoke_share_link. Requires the 'share_link_manage' role permission.

Finalize a presigned upload - the last step after create_document_upload, save_to_dataroom or save_document_version, once the POST to its URL succeeds. Marks the document available and starts processing (PDF conversion, AI classification). Required after those three: until it is called the document is filtered out of every listing, so skipping it strands a file nobody can see. Never after upload_document_base64 or attach_document. They finalize themselves, and a second call runs all of the processing again.

Issue a fresh file URL for a document the reader may see. Called by the preview widget when its file URL expires; not for the model. Pass the `variant` the preview was opened with, so the fresh file_url is the same copy. Named for the widget's call site, which still uses it.