append_to_page_draft
ChatGPTAppend markdown to a page draft. Draft-only and does not publish. Use list_workspaces first and pass the returned workspaceId.
archive_page
ChatGPTArchive exactly one Page as soft lifecycle state. Pages with active descendants cannot be archived and descendants are never archived automatically; authored content, drafts, versions, attachments, and Page identity are preserved, while the Page leaves active Page/tree/search surfaces. Call read_page and read_page_tree first, use the status observed from read_page, and pass the expectedHierarchyRevision from read_page_tree; on conflict inspect current state instead of blindly retrying.
create_child_page
ChatGPTCreate a child page draft under a parent. Draft-only and does not publish. Use list_workspaces first and pass the returned workspaceId.
create_page
ChatGPTCreate a page draft in a workspace. Draft-only and does not publish. Use list_workspaces first and pass the returned workspaceId.
create_page_translation
ChatGPTCreate a DRAFT translation variant for a source page. Optionally provide a localized title; if omitted, the source page title is reused. Draft-only: does not publish or enable a public locale. Requires docs:write. Use list_workspaces first and pass the returned workspaceId.
get_attachment_upload_capabilities
ChatGPTGet backend-authoritative attachment upload policy before prepare_attachment_upload → client raw HTTP upload. Reports whether uploads are enabled and the per-file size limit; it does not upload bytes. upload_attachment validates internally, so it need not be called first.
get_attachment_upload_result
ChatGPTRecover a previously prepared attachment upload when the client binary HTTP request may have completed but its response was unavailable. Returns current backend state and, when completed, canonical Markdown references.
get_document_syntax
ChatGPTRead the bounded, versioned DocRunic Markdown syntax contract. This is global static metadata: it does not inspect workspaces, pages, attachments or private document content.
get_site_analytics
ChatGPTRead bounded public Site analytics for a workspace. Supports 7d, 30d or 90d periods, ISO dates, and day/hour granularity; hourly ranges are limited to 7 UTC days. Returns aggregate completeness states without query text or visitor identifiers.
list_attachments
ChatGPTList attachment metadata for a workspace or page. Use list_workspaces first and pass the returned workspaceId. Omit cursor on first page; only reuse nextCursor from prior response.
list_page_versions
ChatGPTList page version metadata. Use list_workspaces first and pass the returned workspaceId. Omit cursor on first page; only reuse nextCursor from prior response.
list_pages
ChatGPTList page metadata in bounded pages. Use list_workspaces first and pass the returned workspaceId. Omit cursor on first page; only reuse nextCursor from prior response. Omit parentPageId for roots; use page ID for direct children only (not workspace ID).
list_workspaces
ChatGPTList token-allowed workspaces in bounded pages. Omit cursor on first page. Only reuse nextCursor exactly from prior response.
move_page
ChatGPTReorder, reparent, or move a Page to Workspace root (targetParentPageId=null). Call read_page_tree first and pass hierarchyRevision; backend clamps position, moves descendants, and may change slug/path on collision. On hierarchy conflict, the tree changed: re-read and reconsider; do not blindly retry.
prepare_attachment_upload
ChatGPTPrepare a short-lived session for client-accessible local/execution files. Client sends raw bytes to returned DocRunic URL with method+headers; never send local paths or Base64 via MCP. Remote MCP cannot open local paths; model needn't understand images. Use upload_attachment for native handoff.
publish_page_draft
ChatGPTPublishes existing draft only, creates immutable page version, clears draft state, requires docs:publish, and never deletes or archives anything. Use list_workspaces first and pass the returned workspaceId.
read_page
ChatGPTRead current page content with configured byte limits. Use list_workspaces first and pass the returned workspaceId to this tool.
read_page_tree
ChatGPTRead page hierarchy metadata without document content. Use list_workspaces first and pass the returned workspaceId.
read_page_version
ChatGPTRead one page version with configured byte limits. Use list_workspaces first and pass the returned workspaceId.
resolve_internal_page_link
ChatGPTResolve an exact Page ID into the canonical internal Markdown link. Use search_pages or list_pages first, then pass the exact targetPageId and the label to display. This only reads and validates Page metadata; it does not change content or publish.
restore_page
ChatGPTRestore exactly one archived Page to its persisted previous lifecycle status: DRAFT remains DRAFT and PUBLISHED becomes active/published according to normal visibility rules. Call list_pages(includeArchived=true) or read_page(includeArchived=true) plus read_page_tree first, pass expectedStatus=ARCHIVED and expectedHierarchyRevision from read_page_tree. Restore preserves the stored parent, requires it to be active, recalculates path/depth, treats historical position as desired and may clamp it, and returns authoritative position/path; it does not create a PageVersion, publish a draft, restore descendants, choose another parent, or retry conflicts blindly.
restore_page_version
ChatGPTRestore one immutable historical Page version into the current Page draft. Draft-only and does not publish: the historical version remains immutable, the currently published version does not change, and nothing is published automatically. Read the Page/draft state immediately before calling and pass it as expectedDraftState; review the restored draft, then explicitly publish it with publish_page_draft. If a conflict is returned, newer state exists: re-read the Page/draft before deciding what to do and do not retry blindly. Use list_page_versions to choose versionId and optionally read_page_version first.
search_pages
ChatGPTSearch workspace pages for relevant results. Use list_workspaces first and pass the returned workspaceId. Omit cursor on first page; only reuse nextCursor from prior response. Use list_pages for hierarchy traversal; not a full listing tool.
update_page_draft
ChatGPTReplace page draft markdown. Draft-only and does not publish. Use list_workspaces first and pass the returned workspaceId.
update_page_icon
ChatGPTAdd, replace, or remove the current icon on one existing Page. Read the Page first and pass its exact current icon as expectedIcon, including null when no icon is configured. Updates only the target Page, including non-default locale variants; the backend validates catalog, Unicode, attachment ownership, lifecycle, and concurrency rules.
update_page_title
ChatGPTRename one existing active DocRunic Page. Read the Page first and pass its exact current title as expectedTitle. This changes only the Page title: slug/path, hierarchy, locale, translation, content, draft state, published versions, and lifecycle status stay unchanged. Renaming an already-public Page may immediately change the title shown on its public Site.
upload_attachment
ChatGPTUse when the host provides a native file-handoff descriptor for a host-owned/generated file. DocRunic securely obtains and uploads the bytes through the normal attachment pipeline. For client-local files, use prepare_attachment_upload → raw HTTP; the remote MCP server cannot open local paths.