oriiion integration reference

Get the authenticated user's profile and business information.

No parameters.

Update the user's business profile information. All fields are optional — only provided fields will be updated.

Set the user's account type. This affects which social platforms are required during onboarding.

List all social media platforms the user has connected (Facebook, Instagram, LinkedIn, X). Shows page names and usernames.

No parameters.

Check the user's onboarding progress. Returns which steps are completed, which are required, and whether onboarding is fully complete.

No parameters.

Generate onboarding links for the user's remaining steps (payment, Instagram, Facebook, LinkedIn connections). Returns direct links the user can click to complete each step.

No parameters.

Mark the user's onboarding as complete. Only works once all required steps (payment + required social connection) are done — verify with get_onboarding_status first. Call this at the end of a successful onboarding conversation.

No parameters.

Scrape the user's website and auto-extract business name, category and description into their profile (only fills fields that are still empty). Use during onboarding so the user doesn't have to type what their site already says. Show the user what was found and confirm it.

Headline social-media performance for the account over a period: total views, reach, engagements (likes/comments/shares/saves), posts and followers, each with change vs the previous period. Use this to understand overall account health.

Per-platform breakdown of performance (views, reach, engagements, followers, engagement rate) ranked so you can see which platform delivers the most value for this customer. Use it to decide where to focus.

The best-performing published posts by engagement, with their captions, platforms, media type and metrics. Study these to learn what content resonates and create more of what works.

Engagement-weighted best times to post (day + hour, UTC) based on the account's own history, plus concrete upcoming datetimes to schedule at. Use before scheduling to maximise reach.

Follower counts and growth per connected account over a period. Use alongside engagement to judge which platforms are worth investing in.

A synthesized 'what's working' briefing combining the best platform, best posting times, top posts and the engagement mix into actionable guidance for creating better content. Call this before generating new posts to ground them in what performs.

List the user's social media posts. Supports filtering by status and platform. Returns up to 50 posts at a time.

Get a single post by ID, including its carousel media items if any.

Generate new social media posts using AI. Creates posts asynchronously — returns a task ID and group ID to track progress. Use get_generation_status to poll for completion.

Returns immediately — poll get_generation_status for the result.

Check the status of an async post generation batch. Returns 'generating', 'completed', 'failed', or 'partial'.

Update the caption/text of an existing post. Without a platform, the main caption is updated everywhere. With a platform, only that platform's caption variant is changed (other platforms keep theirs).

AI-generate platform-specific caption variants (Facebook, Instagram, LinkedIn, X, etc.) for one post, tailored to each connected platform's style and character limits. The base caption stays unchanged; each platform publishes its own variant.

Add mentions/tags to a post. Instagram: photo tags (usernames) and collaborators (max 3). LinkedIn: mentions as { name, url } pairs — the caption must contain '@<name>' for each one. Applied when the post is published.

Set an automatic first comment for a post (posted right after publishing — commonly used for hashtags or links). Supported on Facebook, Instagram and LinkedIn. Pass an empty string to remove it.

Create two links to a visual page of the user's generated posts. ownerLink = full EDIT powers: edit each platform's caption, AI-edit / replace / add images, and schedule (great when chat can't render images, or as a QR on a phone) — share only with people trusted to edit. reviewLink = read-only EXTERNAL APPROVAL: view every platform's caption, leave comments (a thread) and approve or request changes — safe to share with a client or colleague; their feedback notifies the user and shows in their dashboard Reviews tab. Defaults to the latest generation batch.

Delete a post. Only draft and scheduled posts can be deleted. Published posts cannot be deleted.

Publish a draft post. By default it goes to all the user's configured platforms. Pass platforms to publish to a specific subset only (e.g. only LinkedIn). Only draft posts can be published.

Schedule a draft post for a specific date and time. The post will be automatically published at the scheduled time.

Find the user's social media profiles (Instagram, LinkedIn, Facebook, YouTube, TikTok, X) by harvesting links from their website and web-searching for official profiles. Returns candidates only — confirm with the user, then save with set_social_profiles.

No parameters.

Save the user's confirmed social media profile URLs to their account.

Collect candidate brand images from the user's website, Instagram posts and earlier research (up to 24). Show them to the user to pick from, then save the selection with select_brand_images.

Save the user's chosen brand images as visual references for post generation.

Analyze the user's website (screenshot + AI vision) to extract their brand identity: logo, brand colors and fonts. Takes ~20-30 seconds. Nothing is saved — show the result to the user and persist their confirmed version with set_brand_identity.

Get the user's saved brand identity: logo, colors, fonts, AI-image setting, and content style guide (include/exclude rules).

No parameters.

Save the user's confirmed brand identity. Only provided fields are updated.

Generate a content style guide (tone of voice, dos, don'ts, emoji & hashtag policy) from the user's saved business profile. Nothing is saved — show it to the user, apply their edits, then persist with save_style_guide.

Persist the user's confirmed content style guide. It feeds directly into post generation (include/exclude rules).

Choose whether the user's posts use AI-created images (true) or real photos — their brand images, products and stock (false).

List the user's products (from their connected Shopify store and/or website product feed). Pass `search` to filter by title/description. Each item includes ALL its image URLs in `images` (plus `imageCount`) and a brief description; use get_product for the full untruncated description. Use the returned ids with generate_from_product.

Get full details for one product — complete title, full (untruncated) description and every product image URL. Use before generating so posts reflect the real product info.

Create a new product in oriiion (added to the user's manual product feed). Returns a feed-{id} you can immediately use with generate_from_product. Image URLs are re-hosted on oriiion storage.

Generate social media posts featuring one of the user's products (using its title, description and image). Runs asynchronously — the user is notified when posts are ready.

Returns immediately — poll get_generation_status for the result.

List recent news articles relevant to the user's business/topics. Use the returned article ids with generate_from_news.

Generate thought-leadership posts based on a news article. Omit articleId to auto-pick the most relevant recent article. Runs asynchronously — the user is notified when posts are ready.

Returns immediately — poll get_generation_status for the result.

Store a small file in the user's oriiion account by passing its bytes inline as base64, and get back a permanent URL for the post tools. HARD LIMIT 1MB — the whole file travels as tool-call text, so this is only for things like a logo or a one-page PDF. For anything bigger: POST it to https://oriiion.ai/api/mcp/upload with the header X-API-Key: <this connection's API key> and form field file=@<path> if you can make HTTP requests, otherwise call create_upload_link.

Get the user's own files (photos, video, PDFs, documents) into their oriiion account when you only have them inside this conversation. Returns a short-lived link the user opens in a browser and drops files on; then call list_uploaded_files for the stored URLs, which work with create_post_from_image, create_carousel_from_images, create_post_from_video and create_document_post. IF YOU CAN RUN SHELL COMMANDS, skip this and upload directly instead — it is faster and needs nothing from the user: curl -X POST https://oriiion.ai/api/mcp/upload -H "X-API-Key: <the API key configured for this connection>" -F "file=@/path/to/file" — it returns the stored URL as JSON. Use this link tool when you cannot make HTTP requests or the file only exists as an attachment in the chat.

The URLs of files the user recently uploaded out of band — via an upload link or a direct HTTP upload — newest first. Call this after the user says they have finished uploading. Returns each file's type so you know which post tool to pass it to.

Create posts from an image the user provides (public URL). The image becomes the post's media and AI writes captions for it. Runs asynchronously — the user is notified when posts are ready.

Create a carousel post from 2-10 images (public URLs). Images are cropped to portrait format. Provide either a ready caption or a context for AI to write one. Returns a draft post plus an ownerLink (a page showing every slide where the user can edit, schedule or publish) and a reviewLink for sharing. ALWAYS give the user the ownerLink — it is the only way to see the slides here, since this chat cannot display images.

Create a swipeable document post from a PDF (public URL) — the format LinkedIn shows as a page-by-page carousel. LinkedIn is the ONLY platform that supports documents: if LinkedIn is not connected the tool returns a connect link, which you should give the user. By default the PDF pages are ALSO rendered as JPGs into a separate Instagram carousel post, so the same content goes out on both networks. Provide either a ready caption or a context for AI to write one.

Add an image (public URL) to an existing post as an extra slide, turning it into a carousel (max 10 slides).

AI-edit a post's image with a natural-language instruction (e.g. 'make the background blue', 'remove the text'). Runs asynchronously (~1-2 min) and applies the edited image to the post automatically. Check progress with get_task_status.

Returns immediately — poll get_task_status for the result.

Replace a post's image entirely. Pass imageUrl to swap in an existing image (instant), or prompt to generate a brand-new AI image (async ~1-2 min, applied automatically; check with get_task_status).

Check the status of an async task (image edits, video processing). Returns 'processing', 'completed' or 'failed'. For post generation batches, prefer waiting for the automatic notification instead.

Create posts from a video the user provides (public URL). The video becomes the post's media and AI writes captions based on the topic. Runs asynchronously — the user is notified when posts are ready.

Returns immediately — poll get_task_status for the result.

Add AI-generated burned-in subtitles to a talking video (public URL). Uses the user's saved caption style and brand color unless overridden. The captioned video appears in the user's Todo list when ready (1-3 minutes).

Returns immediately — poll get_task_status for the result.

Turn a script into a talking-head video the user films on their phone. Pass `script` to have them say your exact words (nothing is rewritten — it is only cut into takes), or `topic` to have oriiion write the script. Returns a link that opens a teleprompter + camera on a phone, and shows a QR code to scan when opened on a computer — so it works whichever device this chat is on. The user films each take, then the video renders automatically (1-3 min) and appears ON THAT SAME PAGE, where they can edit the caption, subtitle style, hook and CTA and re-render. This chat cannot be notified when the render finishes — always tell the user the video arrives on that page, and use get_video_shoot_status when they come back and ask.

Returns immediately — poll get_video_shoot_status for the result.

Where a guided video shoot stands: how many takes are filmed, whether the video is still rendering, and the finished video plus its hook and CTA once it is done. Defaults to the user's most recent shoot. Call this when the user asks how their video is coming along — nothing can push that into this chat on its own.

Get the user's prepaid advertising wallet: available balance, reserved funds, service fee, and recent transactions. Amounts are in cents.

No parameters.

List the user's ad campaigns (boosted posts) with status, budget, actual spend, and performance metrics.

No parameters.

Get AI-generated suggestions for which post to boost as a paid ad, with recommended audience, budget, and duration. Takes a few seconds. Use boost_post to launch one.

No parameters.

Launch a paid ad boosting a published post. THIS SPENDS MONEY from the user's prepaid ad wallet (budget + service fee). Always confirm the budget with the user before calling. Use get_boost_recommendations first for suggested settings.

Get oriiion's subscription plans and prices. Works without authentication — useful before the user has an account. To subscribe, the user pays through their personal payment link from get_onboarding_links.

No parameters.

Get the authenticated user's own subscription/contract status: current plan, monthly amount, last and next payment, recent invoices, and Try & Buy trial state (on trial + days remaining). Scoped to the signed-in user only. Invoices are emailed automatically to the invoice email on file; there is no on-demand resend.

No parameters.

Read recent comments on the user's social media posts across platforms. By default returns only unanswered comments (no reply from the account yet). Use reply_to_comment to answer one.

Reply to a comment on one of the user's social media posts. Use the postId, accountId and commentId values exactly as returned by list_comments (these are platform ids, not oriiion post numbers). Always show the user the reply text before sending unless they already approved it.

List the account's comment→DM automations: when someone comments a keyword on a post (or replies to a story), the account automatically DMs them and can post a public reply. Instagram and Facebook only. Also returns the accountId values to use when creating one. Admin-only feature.

No parameters.

Create a comment→DM automation. Once created it is LIVE and will DM real people automatically, so show the user the exact dmMessage, keywords and which account it runs on, and get their approval first. Use an accountId from list_comment_automations. Instagram and Facebook only.

Update a comment→DM automation, or pause/resume it. Passing isActive:false alone pauses it without changing anything else; isActive:true resumes it. Only the fields you pass are changed.

Permanently delete a comment→DM automation. To stop it temporarily, use update_comment_automation with isActive:false instead. Confirm with the user before deleting.

List the account's DM broadcast campaigns with status (draft, scheduled, sending, completed, cancelled), recipient counts and message text. Also returns the accountId values to use when creating one. Admin-only feature.

No parameters.

Create a broadcast DRAFT — a DM sent to many contacts at once. Creating never sends anything: add people with add_broadcast_recipients, then deliver with send_broadcast. Use an accountId from list_broadcasts. Not available on WhatsApp or LinkedIn.

List the contacts a broadcast can reach — people who have messaged the account and are still subscribed. Use this to pick contactId values for add_broadcast_recipients.

Add people to a broadcast draft, or just read who is already on it (call with neither contactIds nor populate to read). Recipients can only ever be ADDED — there is no way to remove one, so trimming an over-filled draft means deleting it and starting over. Adding still sends nothing.

Deliver, schedule or cancel a broadcast. THIS DMs REAL PEOPLE AND CANNOT BE UNDONE. Before calling with action send or schedule you MUST show the user the exact message text and the number of recipients, and get their explicit approval. You must also pass confirmRecipientCount matching the current recipient count (from add_broadcast_recipients) — the call is refused otherwise. action cancel stops a scheduled broadcast and needs no confirmation.

Delete a broadcast and its recipient list. A broadcast that is mid-send must be cancelled first. Confirm with the user before deleting.

List the user's connected photo libraries (Dropbox, OneDrive, Google Drive/Photos, etc.) and which approved folder/album each imports from.

No parameters.

Search the user's own photo library (imported + scraped, vision-tagged) by keyword — e.g. 'coffee', 'team', 'outdoor'. Returns real photo URLs to use with change_image / add_image_to_post.

Given a post ID, return the best-matching real photos from the user's own library for that post's topic (ranked). Ideal for AI-OFF customers — pass a chosen URL to change_image to set it.