copy_folder
ChatGPTCopies a folder with everything in it — its subfolders and the emails or templates they hold, recursively — and returns the new folder's id and name plus how many items were copied. The source stays untouched. With targetParentFolderId the copy goes inside that folder (same kind, email or template, as the copied one); without it the copy lands in the root of the folder's own project. Same rules as the Stripo UI: the copy keeps the folder's name (numbered when a sibling already has it) and the project's folders limit and emails-and-templates quota apply — when there is no room the call fails with a message saying so. All or nothing: a failed copy leaves nothing behind. Not idempotent: a retry makes another copy; after a timeout check find_folders(nameContains=...) first. See stripo://guide/start-here.
create_email
ChatGPTCreates a new, editable email in a Stripo project and returns its emailId; the content tools all operate on a letter that already exists. Every parameter is optional: with no source the letter is blank, with no folderId it lands in the project root, and with no projectId in the project the user worked in last — the response reports the projectId and the name actually used. sourceTemplateId builds the email from a template and sourceEmailId copies an existing one; the two are mutually exclusive. The response also carries an 'instruction' field naming the follow-up call for this deployment. Not idempotent: each call creates another email, and an abandoned one stays in the project until it is deleted. Retrying after a timeout can create a second email; find_content(nameContains=...) shows whether the first call already succeeded. See stripo://guide/start-here.
create_folder
ChatGPTCreates a folder for emails or templates and returns its folderId and the name actually stored. With parentFolderId the folder is nested inside that folder and takes its project and kind (contentType is ignored); without it the folder lands in a project root and contentType says which kind. Same rules as the Stripo UI: a name already taken gets a numeric suffix rather than an error, and the project's limit on the number of folders applies — when it is reached the call fails with a message saying so. Not idempotent: a retry creates another, numbered folder; find_folders(nameContains=...) shows whether the first call already succeeded. See stripo://guide/start-here.
delete_content
ChatGPTSoft-deletes one or more emails, templates or modules by type and ids. The call requires confirm=true; dryRun=true previews it instead, reporting what would be deleted without deleting anything, and takes precedence when both are set. Returns the requested count plus deleted[] (ids soft-deleted by this call, or that would be on a dry run) and denied[] (ids not deleted, each with a reason: NOT_FOUND or FORBIDDEN). Deleted items stop appearing in find_content and get_content and remain recoverable from the trash bin. See stripo://guide/access.
edit_image
ChatGPTStarts editing ONE existing image by instruction, optionally with reference images, for an email or template and returns a jobId. The source is never modified: a new PNG is created in the letter's gallery and marked as AI-generated. Same access, quota and polling rules as generate_image: every call spends quota and a repeated call is a new job. See stripo://guide/recipes.
export_to_html
ChatGPTExports one email or template as a ready-to-send html document compiled by the Stripo export pipeline (styles inlined, dev markup stripped) — unlike get_content, which returns the raw stored html. The compiled HTML is never returned inline: the response carries htmlUrl, a temporary presigned link to the compiled document, valid until htmlUrlExpiresAt; ampHtmlUrl and ampHtmlUrlExpiresAt work the same way. mergeTags substitutes placeholder values before export; placeholders left unreplaced are listed in warnings[] rather than raised as errors. Each call consumes one export from the workspace quota (status=QUOTA_EXCEEDED once it is spent). Modules are not supported. A missing or inaccessible id yields status=NOT_FOUND rather than an error. See stripo://guide/access.
find_activity
ChatGPTSearches create/update/delete/move/restore events across emails, templates, modules and folders in the workspaces the authenticated user can access. ACL matches find_content: results are restricted to the projects the user has permission on. Filters: type (EMAIL|TEMPLATE|MODULE|FOLDER), operation (CREATE|UPDATE|DELETE|MOVE|RESTORE), actorId, entityId, workspaceId, scope (current | all), projectId (single-project narrowing; must be in the accessible scope), folderId, occurredAfter/occurredBefore (epoch millis). Paginated via limit (default 25, max 100) and offset. When activity logging is unavailable or no projects are accessible, the events list is empty and a note explains the state. See stripo://guide/recipes.
find_content
ChatGPTSearches or counts emails, templates and modules by name, body or metadata; returns typed, named hits with total and byType. Returns at most limit items (default 25, max 100) starting at offset; total is the full match count, and hasMore=true means further pages remain at nextOffset. countOnly and groupBy return counts or a breakdown without listing items. Drafts are excluded by default; the draft flag includes or isolates them. See stripo://guide/recipes.
find_folders
ChatGPTLists and searches the folders the authenticated user can access, as a paginated set. Each folder carries its item count (non-draft, original-language emails or templates directly in it), an empty flag, its folderType (EMAIL or TEMPLATE), its projectId/projectName and its parentFolderId (absent for root folders), which together describe the folder tree. Searches every accessible workspace by default; scope=current restricts to the current workspace, and workspaceId or projectId narrow it further. nameContains filters by name; limit/offset page the result (the response carries total, returned, hasMore and nextOffset). See stripo://guide/start-here.
find_projects
ChatGPTFinds the projects the authenticated user can access, with per-project detail: workspace and group, archived flag, creator, created/updated timestamps, last content activity timestamp (lastActivityAt, epoch-ms; null when no activity events are recorded), content counts (emails, templates, modules) and the user's effective role, plus a roles legend. Filters: name substring, workspace, creator, created/updated epoch-ms ranges. Paginated via offset/limit. scope=current (default) stays in the current workspace; scope=all spans every accessible workspace. A single projectId returns that project with extra industry and logo detail; an out-of-scope projectId is ignored and the full reachable set is returned with an explanatory note. See stripo://guide/start-here.
generate_image
ChatGPTStarts generating ONE new image for an email or template from a text prompt, optionally guided by reference images, with Stripo AI (OpenAI gpt-image) and returns a jobId. The result is a PNG hosted in the letter's image gallery and marked as AI-generated (C2PA). Requires write access to the letter and the AI image generation right in its project. Every call spends the project's AI image quota, including results you discard, and a repeated call starts a new job, so do not retry a failed job with the same prompt blindly. Poll get_image_job(jobId) every pollAfterSeconds until completed; generation usually takes 20–90 seconds. The image is not placed into the letter: put image.url into an image block's src through the document-state round-trip. See stripo://guide/recipes.
get_business_profile
ChatGPTRead a project's Business Profile. Returns businessProfileId and businessProfile; if no Business Profile exists, businessProfileId is null and businessProfile contains defaults. Use after changes to check the saved result. See stripo://guide/recipes.
get_content
ChatGPTReturns the full detail of one email, template or module by type and id: name, project/folder, author and editor (with names), timestamps, draft flag, language, and for emails the subjectLine/preheader derived from the html. HTML is never returned inline: with includeHtml=true the response carries htmlUrl, a temporary presigned link to the HTML document, and htmlUrlExpiresAt, the instant that link stops working. Read-only — it never modifies the content. A missing or inaccessible id yields status=NOT_FOUND rather than an error. See stripo://guide/access.
get_document_state
ChatGPTReturns the current document state (content JSON) of one letter, for inspection or for a get -> modify -> set_document_state round-trip. The JSON is not returned inline: the response carries downloadUrl, a temporary presigned link to the document, and expiresAt, the instant it stops working. Read access is checked first — without it the status is ACCESS_DENIED and no URL is returned. See stripo://guide/recipes.
get_document_state_schema
ChatGPTReturns the JSON Schema every letter's document state follows. It takes no arguments: the schema is global, not the schema of one letter. The schema is not returned inline: the response carries downloadUrl, a temporary presigned link to the schema document, and expiresAt, the instant it stops working. set_document_state validates its input against this schema. See stripo://guide/recipes.
get_image_job
ChatGPTReturns the state of an image job. running: call again after pollAfterSeconds. completed: image.url is the hosted PNG (permanent, in the letter's gallery) with width, height and bytes; calling again returns the same image, nothing is hosted twice. failed: error.code explains — quota_exceeded (details.reason SUBSCRIPTION_LIMITS_REACHED is the plan limit for this billing period, DAILY_LIMITS_REACHED a global daily cap; neither is fixed by retrying) or generation_failed (the provider rejected or failed; change the prompt or references, a new call spends quota). expired: the result was not collected within 5 minutes of completion; start a new job. See stripo://guide/recipes.
get_screenshot
ChatGPTReturns temporary PNG download links for the export-ready view of the current coediting state of an email or template, including unsaved changes and the project's export settings. mode is required and accepts DESKTOP, MOBILE, or BOTH (case-insensitive). BOTH returns DESKTOP then MOBILE; if only one renders, status is PARTIAL. Access is checked before rendering; download every returned file before its link expires. See stripo://guide/recipes.
get_server_time
ChatGPTReturns the server's current time in UTC: epoch millis, ISO instant, today's date and day of week. It is the reference instant for the relative date windows (yesterday, last 30 days, ...) and epoch-ms filters that the other Stripo tools accept. See stripo://guide/dates.
list_members
ChatGPTLists the members of a workspace with their workspace role and an access breakdown (access.scope is WORKSPACE, GROUPS or PROJECTS; group and project entries carry their own roles). Defaults to the authenticated user's current workspace; workspaceId selects another accessible one. projectId keeps only members that can access that project — each then carries projectAccess with effectiveRole and grantedVia (WORKSPACE, GROUP or PROJECT). Suspended members and pending invites are excluded. The roles legend describes every referenced role. limit/offset page the result (the response carries total, returned, hasMore and nextOffset). Requires permission to view workspace members (owners and admins have it by default); otherwise the call is denied. See stripo://guide/access.
patch_business_profile
ChatGPTUpdate selected Business Profile fields with a JSON object. Omitted fields stay unchanged; supplied arrays replace the whole list, and [] clears it. Creates a Business Profile if none exists. Use replace_business_profile to replace the whole Business Profile. Read get_business_profile after writing or before retrying an unconfirmed write. See stripo://guide/recipes.
prepare_business_profile_upload
ChatGPTPrepare a JSON file upload to replace a project's Business Profile. Returns an uploadUrl, an uploadId and upload instructions. Follow the instructions to upload the file, then call replace_business_profile with the same projectId and uploadId. See stripo://guide/recipes.
prepare_document_state_upload
ChatGPTStep 1 of writing a letter's content JSON: returns a single-use uploadId and a temporary presigned S3 PUT url you upload the new document state to yourself. The JSON is never passed inline. Pair it with get_document_state for a programmatic round-trip, and validate what you upload against get_document_state_schema first. Write access is checked here, so ACCESS_DENIED means no ticket was minted and nothing can be written. Follow the returned instruction verbatim: upload the file, then call set_document_state with the uploadId — never the url, which is a short-lived credential. The url expires in minutes. One ticket holds one file: to keep concurrent edits by others, call this twice and upload the unmodified get_document_state JSON to the second ticket, passed as baseUploadId. The uploadId is spent once set_document_state has read the file; a call made before the upload finished spends nothing. See stripo://guide/recipes.
prepare_image_upload
ChatGPTStep 1 of hosting an image the user supplied (a chat attachment or a local file) in the image gallery of an email or template: returns a single-use uploadSessionId and, in uploads[0], a temporary presigned S3 PUT uploadUrl with the accepted contentTypes and maxBytes. The file is uploaded to that url by the client itself; image bytes, Base64 and local paths are never tool arguments. PNG, JPEG and GIF are accepted; what counts is the actual bytes, not the file name. PNG and GIF dimensions are also limited (4000x4000 by default); the instruction gives the current limit. Requires write access to the letter and the right to change its images; no AI is involved and no AI image quota is spent. The returned instruction names the exact upload command and the upload_image call that follows. The url is a short-lived credential and expires at expiresAt; upload_image takes the uploadSessionId, not the url. An image that already has a public https URL needs no upload. See stripo://guide/recipes.
replace_business_profile
ChatGPTFully replace a project's Business Profile using an uploadId from prepare_business_profile_upload. Upload the JSON file first and use the same projectId. For a regular Business Profile, the JSON must contain a root brand object; omit website. Supply the source website only when importing already extracted data. Replaces all current Business Profile data and may overwrite concurrent changes. Each uploadId is single-use. Read get_business_profile after writing or before retrying an unconfirmed write. See stripo://guide/recipes.
set_document_state
ChatGPTApplies a document state previously uploaded via prepare_document_state_upload for the same id/type, identified by uploadId; a URL or inline JSON is not accepted here. Passing baseUploadId — a second ticket carrying the unmodified get_document_state JSON — turns the write into a delta on top of the letter's current state, so concurrent edits by others survive; without it, the write replaces the whole document. Tickets are single-use. UPLOAD_NOT_FOUND names the empty ticket in missingUploadId, and every non-OK status carries an instruction field naming the next call. MERGE_BROKEN carries a code and, for VALIDATION_ERROR, details.errors[{path, message}]. REVISION_CONFLICT means the letter changed while the write was running and fresh tickets are needed. WRITE_UNCONFIRMED means the outcome is unknown until get_document_state confirms it. See stripo://guide/recipes.
show_content_preview
ChatGPTReturns the native shared-preview and editor links for the current email or template. Supported hosts display an MCP App as the final presentation step after a confirmed save and required checks, or an empty create_email with editorModelReady=true. On OK the editor link is the user-facing result; the App displays the preview link, or a desktop screenshot when the workspace plan does not include shared preview. Failure does not undo a save and requires no repeated create/write. No export quota is consumed. See stripo://guide/recipes.
update_content
ChatGPTRenames an email or template and/or moves it between the folders of its project — the way to tidy up after create_email, which can only put a new letter in one folder under one name. Pass at least one of name and folderId; passing only one is fine, omitting both is an error. The new name is stored exactly as given: names need not be unique in a project, so nothing is renumbered and no conflict is reported. The folder must belong to the same project — content cannot be moved to another project through this API. Nothing else changes: the content itself, its sharing settings and its editor model are untouched; the item's language versions follow it into the new folder, as they do in the UI. The response reports the state after the change plus renamed/moved, so a no-op is visible. Refused while the item is open in the editor, by anyone. Safe to retry: the same call twice leaves the same result. See stripo://guide/access.
upload_image
ChatGPTStep 2: hosts the file uploaded through prepare_image_upload in the same email or template gallery. Success is data{url, name, contentType, bytes}: data.url is the permanent public URL for an image block's src, set through the document-state round-trip; a PNG or JPEG url also serves as imageUrl or referenceImageUrls of the AI image tools. This call does not insert the image. JPEG and GIF are stored byte for byte, keeping animation; PNG keeps pixels and transparency but loses metadata. PNG and GIF dimensions default to at most 4000x4000. Failure is {error_code, reason, instruction}: upload_session_not_found (not uploaded, expired, used, or another letter/user), file_too_large (byte size or dimensions), unsupported_format (not PNG/JPEG/GIF or an invalid header), hosting_rejected (refused, nothing stored), hosting_failed (outcome unknown). The session is single-use: once the file is read it is spent whatever the answer; a call made before the upload finishes spends nothing. See stripo://guide/recipes.
whoami
ChatGPTReturns the authenticated user's identity and access: profile (userId, name, email, avatarUrl); every accessible workspace with the user's role, access scope and content counts — emails, templates and modules at the workspace, group and project level, plus the accessible project count per workspace; and a roles legend describing what each role can do. Identity comes from the authenticated session, so the tool takes no parameters. See stripo://guide/start-here.