hoja_get_upload_status
ChatGPTPoll the status of a browser upload handoff and retrieve its stored file ID after completion.
hoja_prepare_file_upload
ChatGPTCreate a short-lived HOJA upload ticket, then send the file with hoja_upload_file_bytes. For a file already in this chat (such as an image you generated) use hoja_upload_chat_file instead - no ticket needed. The browser handoff in the response is only for files the person must choose themselves - never ask the user to upload a file you already have. Remote URL uploads are not supported.
hoja_upload_chat_file
ChatGPTSend a file from this chat (for example an image you just generated) to HOJA as a file - no base64. Pass the image itself as file and the upload purpose. Returns a stored file id to pass to the tool that consumes it, such as attaching a storyboard scene image. Prefer this over hoja_upload_file_bytes.
hoja_upload_file_bytes
ChatGPTSend the bytes of a file you created (for example an image you just generated) to HOJA, using a ticket from hoja_prepare_file_upload. Returns a stored file id to pass to the tool that consumes it, such as attaching a storyboard scene image. Use this instead of asking the user to download and re-upload the file. contentBase64 must be the base64 of the original file produced in code - never typed, shortened or re-compressed by hand.
me_leave_balance
ChatGPTGet the current user's own leave balance by leave type: entitled, used, pending and remaining days.
wAttachMarketingVideoSceneImage
ChatGPTAttach an image the user supplied (their own photo or artwork) as a scene's storyboard image. For a series template video do not draw scene images in the chat - HOJA draws them with GPT Image (wStartMarketingVideoStoryboardImages). Upload the file first with hoja_prepare_file_upload and pass the resulting stored file id. This revokes any previous image approval and marks an existing scene video as stale. For a series template video HOJA then checks the image against that scene's look and subject (keyframe QC, a small credit charge per attached image); about a minute later read wListMarketingVideoScenes and redo any scene whose imageQualityVerdict is NEEDS_REVIEW or NOT_RECOMMENDED, following imageQualityFindings.
wConfirmMarketingResearchDossier
ChatGPTConfirm which verified research facts become the planning evidence of the video (replaces the brief's grounding facts). The first call only returns a summary and a confirmationId; nothing changes yet. Show the summary to the user, and only after they agree call this tool again with the same arguments plus that confirmationId.
wCreateCustomer
ChatGPTCreate a new customer record in HOJA. HOJA generates the customer code.
wCreateCustomerContact
ChatGPTAdd a contact person (name, phone, email, position) to an existing customer.
wCreateMarketingVideoProjectFromTemplate
ChatGPTCreate an AI video project from a series template for one topic (template code, topic, brand, brand usage NONE or CONTEXT, look preset, and the length the user stated in this conversation in seconds as the planning target - if they have not said it in this conversation, ask them first as an open question, never as fixed options; a length remembered from an earlier conversation does not count). For a team template pass customTemplateId (and its base template code). The template fixes the story beat order, the look and camera per beat and external narration; the number of scenes is decided by the storyboard you write, and each scene's length by its narration.
wCreateMarketingVideoSeriesEpisode
ChatGPTCreate one episode in a template video series plan queue from a topic. It creates a template video project with the series' template, length, look, language and aspect ratio (free, no AI call); then research, confirm facts and generate the storyboard for that project.
wDeleteCustomerContact
ChatGPTDelete a customer contact person. This cannot be undone, so it needs the user's approval first. The first call only returns a summary and a confirmationId; nothing changes yet. Show the summary to the user, and only after they agree call this tool again with the same arguments plus that confirmationId.
wDeleteMarketingVideoScenes
ChatGPTDelete one or more storyboard scenes from the same storyboard in one all-or-nothing action. First list the scenes, then pass that storyboardId and every selected scene id with its current expectedRowVersion. If one scene is missing, already deleted, from another storyboard, or changed elsewhere, none are deleted. The paid scene takes remain recoverable with the deleted scenes. The first call only returns a summary and a confirmationId; nothing changes yet. Show the summary to the user, and only after they agree call this tool again with the same arguments plus that confirmationId.
wEstimateMarketingVideoCredits
ChatGPTEstimate the AI credit cost of a video generation stage and return the frozen quote hash required to start it. Always call this before starting storyboard image generation.
wEstimateShopSet
ChatGPTGet the credit estimate for the next paid shop-set step (stage PLAN, GENERATE, REGENERATE_SLOT with slotId, or RESEARCH_GROUP) before running it. Free; nothing changes.
wGetAttendanceLeaveSummary
ChatGPTGet an attendance and leave summary for a period, for the user or for the team they can see.
wGetCustomerSummary
ChatGPTGet one customer's profile summary (contact details, grade, recent activity and spending) by customer ID.
wGetInventoryStatusStats
ChatGPTGet an overview of current stock levels, optionally for one branch or product.
wGetMarketingResearchDossier
ChatGPTGet the latest official-source research dossier of a template video project: each fact card with its source URL, quote and verification status (VERIFIED, VALUE_ONLY, UNVERIFIABLE, MISMATCH, BROKEN), plus the research slots the template asks for (researchSlots[].key is the value to use as slot when submitting facts). Call this before submitting facts.
wGetMarketingVideoProject
ChatGPTGet one AI video project in full detail including every scene (narration, prompts, story beat, timing, image/video approval status). Call this before editing a storyboard.
wGetMarketingVideoSeriesTemplate
ChatGPTGet one team video series template with its editable settings (look, style block, beat weights, camera per beat, narration speed, music, accent, topic and source policy), rule issues and pilot status.
wGetSalesOrder
ChatGPTGet one sales order with its items, amounts and status by order ID.
wListAttendanceEntries
ChatGPTList check-in and check-out records, optionally filtered by date range, branch, department or employee.
wListCatalogProducts
ChatGPTList products and services in the catalog with name, selling price and status.
wListCustomerContacts
ChatGPTList the contact people saved for one customer.
wListCustomers
ChatGPTList customers in the user's HOJA workspace, with optional filters such as name search, customer type, grade, assigned staff or industry.
wListInventoryAlerts
ChatGPTList inventory alerts such as low-stock or out-of-stock products.
wListInventoryMovements
ChatGPTList stock in/out movements, optionally filtered by product, movement type or date range.
wListLeaveRequests
ChatGPTList leave requests, optionally filtered by date range, leave type, status or employee.
wListMarketingContentDrafts
ChatGPTList marketing content drafts, optionally filtered by status or period.
wListMarketingVideoProjects
ChatGPTList AI video projects for this tenant (title, status, aspect ratio, story format, duration).
wListMarketingVideoScenes
ChatGPTList the scenes of an AI video project with scene number, title, narration, image prompt, motion prompt, story beat, image/video approval status, the aspect ratio each scene image must use, and for series template videos the keyframe check of the attached image (imageQualityVerdict: RECOMMENDED / NEEDS_REVIEW / NOT_RECOMMENDED, null while the check is still running; imageQualityFindings says what is wrong). Pass pendingImageOnly true to get only the scenes that still need an image - an empty result is your proof the work is finished. For a series template video HOJA draws the scene images itself with GPT Image (estimate with stage STORYBOARD_IMAGES and includeTargetTypes ['SCENE_IMAGE'], then wStartMarketingVideoStoryboardImages after the user agrees) - do not draw them in the chat. Call this first to find which scene to edit, and to read each scene's image check after HOJA has drawn it.
wListMarketingVideoSeriesEpisodes
ChatGPTList the plan queue (episodes) of a template video series automation in queue order, with each episode's status (DRAFT, READY, APPROVED, IN_PRODUCTION, PRODUCED, PUBLISHED, ON_HOLD, FAILED), readiness checks and the queue supply summary. The list is the work order: finish one episode to READY, then read the list again.
wListMarketingVideoSeriesTemplates
ChatGPTList video series templates. Before making a template video, ask the user its length as an open question (e.g. 'about 1 minute'), never as fixed options like 20/30/60 s.
wListPayments
ChatGPTList received payments, optionally filtered by customer, method, status or date range.
wListQuotes
ChatGPTList sales quotes, optionally filtered by customer, owner, status or a search term.
wListReceivables
ChatGPTList outstanding customer receivables, optionally filtered by customer, status, aging bucket or date range.
wListSalesOpportunities
ChatGPTList sales opportunities in the pipeline, optionally filtered by stage, customer, owner or date range.
wListSalesOrders
ChatGPTList sales orders, optionally filtered by customer, status or a search term.
wMarkMarketingVideoSeriesEpisodeReady
ChatGPTCheck whether an episode is ready (free): confirmed research with every required fact, storyboard passing the template contract, length within the channels' limit, unique topic, publish title and caption, and (keyframe level) approved keyframes. Passing sets READY; failing checks are returned so you can fix them and check again. Approval happens in the HOJA web app.
wPlanShopSet
ChatGPTPlan a shop set (slots with purpose and fact-backed claims) from the research or the brand data. Costs credits: first call wEstimateShopSet with stage PLAN and show the estimate to the user. The first call only returns a summary and a confirmationId; nothing changes yet. Show the summary to the user, and only after they agree call this tool again with the same arguments plus that confirmationId.
wRegenerateShopSetSlot
ChatGPTGenerate one shop-set slot again. Costs credits: first call wEstimateShopSet with stage REGENERATE_SLOT and that slotId, and show the estimate to the user. The first call only returns a summary and a confirmationId; nothing changes yet. Show the summary to the user, and only after they agree call this tool again with the same arguments plus that confirmationId.
wReorderMarketingVideoScenes
ChatGPTReorder storyboard scenes.
wRequestMarketingResearchDossier
ChatGPTStart official-source web research for a template video project's topic (paid AI call, runs in the background for a few minutes).
wRequestMarketingVideoCompliancePreflight
ChatGPTRun the compliance preflight of a video project (free, no credits). Pass intendedScopes ['SOCIAL_ORGANIC'] for normal social posting (add 'PAID_ADS' only when the user says it will run as an ad). HOJA's paid stages - storyboard images and scene videos - refuse to start unless a fresh preflight passed after the last edit, so call it right before wStartMarketingVideoStoryboardImages or scene video generation; if it reports blocking issues, show them to the user and stop.
wRequestMarketingVideoSceneImage
ChatGPTRedraw the storyboard image of one scene using a credit quote from the estimate tool. This spends credits, so it needs the user's approval first. The first call only returns a summary and a confirmationId; nothing changes yet. Show the summary to the user, and only after they agree call this tool again with the same arguments plus that confirmationId.
wStartMarketingVideoStoryboardImages
ChatGPTStart STAGE 1 — HOJA draws one storyboard image per scene from each scene's image prompt (leave the image model empty for HOJA's default, GPT Image) and checks each picture. Requires a frozen quote: call wEstimateMarketingVideoCredits with stage STORYBOARD_IMAGES and includeTargetTypes ['SCENE_IMAGE'] first, show the user the credits, run the free wRequestMarketingVideoCompliancePreflight right before (it refuses to start without a fresh preflight), and pass back its quoteHash once they agree. Consumes AI credits. The first call only returns a summary and a confirmationId; nothing changes yet. Show the summary to the user, and only after they agree call this tool again with the same arguments plus that confirmationId.
wStartShopSetGeneration
ChatGPTGenerate every included slot of a planned shop set (copy, images, product video). Costs credits: first call wEstimateShopSet with stage GENERATE and show the estimate to the user. The first call only returns a summary and a confirmationId; nothing changes yet. Show the summary to the user, and only after they agree call this tool again with the same arguments plus that confirmationId.
wStartShopSetResearch
ChatGPTResearch top-selling listings for one shop-set research group. Costs credits: first call wEstimateShopSet with stage RESEARCH_GROUP and show the estimate to the user. The first call only returns a summary and a confirmationId; nothing changes yet. Show the summary to the user, and only after they agree call this tool again with the same arguments plus that confirmationId.
wSubmitMarketingResearchFacts
ChatGPTSubmit facts you researched yourself (e.g. with web search) for a template video project: one fact per research slot (slot must be a researchSlots[].key from wGetMarketingResearchDossier) with the claim, value, unit, the exact quote as it appears on the page, the source URL and publisher. If any fact is malformed the whole submission is rejected. HOJA fetches every source page and checks the quote; only VERIFIED/VALUE_ONLY facts can be confirmed. No HOJA credits. The first call only returns a summary and a confirmationId; nothing changes yet. Show the summary to the user, and only after they agree call this tool again with the same arguments plus that confirmationId.
wUpdateCustomer
ChatGPTUpdate an existing customer's details such as name, contact information, address or assigned staff.
wUpdateCustomerContact
ChatGPTUpdate the details of an existing customer contact person.
wUpdateMarketingVideoProject
ChatGPTUpdate an AI video project's settings (title, brief, target duration, story format, narration language, voice).
wUpsertMarketingVideoScene
ChatGPTCreate or update one storyboard scene — this is the main tool for writing storyboard data from chat (narration text, image prompt, motion prompt, on-screen text, duration, story beat, visualization type). To set timed subtitles for the scene, send subtitleCuesJson as a JSON array of {startMs, endMs, text} where the times are milliseconds measured from the start of THIS scene, in order and never overlapping; send [] to clear them and go back to subtitles timed automatically from the narration line. Read the scene first and keep any cues already there unless the person asked you to change them.
wValidateMarketingVideoStoryboard
ChatGPTValidate the current storyboard of a video project with the same checks HOJA uses before paid generation (template skeleton, camera grammar, forbidden text in image prompts, evidence references, narrative rules). Call it after writing scenes and fix every issue. When it passes, follow the nextStep it returns: HOJA draws the scene images with GPT Image after the user agrees to the quote. No credits.