Claude (claude.ai and the desktop app)
Settings → Connectors → Add custom connector. Paste the URL, then sign in to Lumen Brain when the connector asks.
MCP · version 0.17.0 · 75 tools
Connect Lumen Brain to Claude, ChatGPT, Cursor or Codex and the assistant works in your account: it generates images and clips, files them in the Brain, searches and reads the documents you keep there, builds and edits designs in the Design Studio, runs the AI Post Editor, reads scripts aloud with ElevenLabs, and schedules posts to your social accounts. Your credits, your gallery, your folders.
One streamable-HTTP server, signed in with OAuth. The client opens Lumen Brain's sign-in once; from then on every tool runs as you.
https://lumenbrain.app/mcpSettings → Connectors → Add custom connector. Paste the URL, then sign in to Lumen Brain when the connector asks.
One command in the terminal, then /mcp to sign in.
claude mcp add --transport http lumen-brain https://lumenbrain.app/mcp
Add the server to ~/.cursor/mcp.json (or the project's .cursor/mcp.json) and sign in from the MCP settings.
{
"mcpServers": {
"lumen-brain": { "url": "https://lumenbrain.app/mcp" }
}
}Settings → Connectors → Create (developer mode). Paste the URL; the sign-in opens on Lumen Brain.
Add it from the terminal, then sign in when Codex asks.
codex mcp add lumen-brain --url https://lumenbrain.app/mcp
Credits
Generation tools spend your Lumen credits at the prices their descriptions state; listing and reading tools are free. get_credits shows the balance.
Outputs
Every image, clip, design render and voiceover lands in your gallery with a public link, and can be filed into a Brain folder server-side — nothing travels through the chat.
Views in the chat
Hosts that support MCP Apps show Lumen Brain's own views: the generation view for images and clips, the design view that paints renders right in the conversation.
Skills and notes
Folder notes and your saved skills are the standing context: an assistant reads them before it works and keeps what it learns there.
Guide
Everything the studio does — templates, slots, layers, layout, renders, AI writing, projects, batches, conversations, sharing — is a tool. A typical job reads top to bottom.
Step 1
list_design_templates lists the presets everyone has (starter:<id>) and the user's own templates with their slots — the text and picture layers a template fills in — and each preset's own instructions. list_designs reads the library; list_design_sizes the page sizes, list_design_fonts the fonts.
Step 2
get_design with include_layers returns the layers as a table: id, name, type, box, the words, the picture, the fill — bottom to top. Those ids (or names, or slot keys) are what every edit names.
Step 3
create_design pours a value per slot into a copy of the template: text as it is (within max_chars), a picture as a Brain asset id, a gallery generation id, an https link or "empty". pictures: "web" finds a picture for every empty frame on the web.
Step 4
edit_design applies a list of the editor's own moves in one call — add_text, add_picture, add_frame, add_shape (27 kinds: polygons, stars, arrows, symbols, callouts, flowchart), add_connector (a line that joins two layers and follows them when they move), add_drawing, add_gradient, set, move, resize, remove, duplicate, reorder, align, set_background, remove_background_picture, resize_page, set_canvas (a page or an infinite canvas), set_instructions, fill_frames — and saves once, with the layers back as a table and a warning when copy prints over copy.
Step 5
check_design_layout measures a design (or a document not yet saved) for copy that outgrows its box or prints over another layer, and with repair: true moves things apart the way the studio does; save: true writes the repaired document back.
Step 6
remove_background takes a Brain picture, a gallery output or a public link and files the subject as a transparent PNG beside the original, ready to drop into a frame. Hair and soft edges survive it. Free, within the day's allowance the plan carries, two at a time; profile: "fine" runs the matting model again when an edge comes back rough, and counts as three.
Step 7
compress_image makes a smaller copy of a Brain picture, a gallery output or a public link at high quality — the same pixels, re-encoded — and files it beside the original with "compressed" after its name. quality: "high" (the default), "balanced" or "small"; format: "auto" keeps a JPEG a JPEG and turns the rest into WebP, transparency kept. Free.
Step 8
render_design queues a picture — painting needs a browser: in a host that shows Lumen Brain's views it is painted right in the chat, otherwise the moment the user has Lumen Brain open. get_design_render returns the public link. review_design_renders has Claude look at the finished pictures and say what is wrong.
Step 9
generate_design writes a complete design from a brief at a size or preset; revise_design applies one instruction; generate_design_variants writes N versions of a template from a brief or one row per line. research: true on any of them searches the web first and writes from dated facts (research_topic on its own does the search), find_web_pictures files pictures of a subject into the Brain.
Step 10
create_design_project builds a multi-page document (a carousel, a campaign); render_design_project paints every page. generate_design_variants with batch: true keeps the run on the AI Post Editor's Bulk tab (list_bulk_batches); generate_design with conversation: true keeps the brief on its Create tab, and continue_design_conversation takes it further turn by turn. share_design (admins) publishes a design as a template for everyone.
Guide
The voice studio's catalogue, samples, readings, voices of your own, video voice changes and transcripts, as ten tools.
Step 1
list_voices lists the models — Eleven v4 (the newest and the default), Eleven v4 Turbo (cheaper, near-instant) and Eleven v3, all ElevenLabs — with their price per started 1,000 characters and every preset voice, with a line on how it sounds and the words it is found by: old, deep, warm, young, british, conversational. Signed in, it adds the user's own voices and marks their favourites; search and favorites_only narrow the list.
Step 2
voice_preview returns an MP3 of one voice reading a fixed line — made once and kept, so it is free and immediate for a voice anyone has heard before. set_voice_favorite hearts the ones worth keeping at hand.
Step 3
generate_speech reads up to 5,000 characters a call on v4 and Turbo (6,000 on v3) and saves the MP3 to the gallery with a public link. Direction goes in square brackets in the text, in English whatever the script's language — [whispers], [excited], [sighs]; stability, similarity and language tune the read. The generation id is a voice track for the video models that take one, and save_output_to_brain files it into a Brain folder.
Step 4
clone_voice makes a voice from a minute of someone's speech (their consent confirmed first), design_voice makes a new one from a description of how it should sound, and add_voice_by_id adds any voice from ElevenLabs' Voice Library by its ID. Each is then custom:<id> in the voice tools — change_video_voice takes ElevenLabs voices only, not a MiniMax clone or a designed voice.
Step 5
change_video_voice re-voices a gallery clip, a Brain video or a public link with any ElevenLabs voice — timing and emotion kept, the music and effects mixed back under it — and saves a new clip to the gallery. It returns at once; get_voice_change follows it. 12 credits a minute.
Step 6
transcribe_audio turns a recording into text — a gallery output by id (a clip, a voiceover), a public url, or bytes — and keeps the transcript in the gallery. 1 credit per started five minutes; pass duration_seconds when it is known.
Guide
Office, Adobe and OpenDocument files, PDFs, e-books and plain text are read as they arrive, so an assistant can search inside them and read them over MCP.
Step 1
Word, Excel, PowerPoint, Outlook, Visio, PDF, Photoshop, Illustrator, InDesign, XD and EPS files, OpenDocument, EPUB, RTF, CSV, text, Markdown, JSON, HTML and XML go into the Brain like any file: from the Brain page, through media_upload_widget in the chat, as base64 with save_media, or from a public link with save_media_from_url. Each is read as it arrives.
Step 2
list_media with kind: "document" lists a folder's documents with their format and page count. search_brain searches inside every document at once — accents optional, Hungarian and English word forms understood, "quoted phrases", or and -word taken — and names up to three pages, slides or sheets of each where the words are, each with a snippet and a from. Several words are found only where they appear together — on one page, slide or sheet, or in one stretch of a long text — so start with one distinctive word; whole words that find nothing are tried again as the starts of words.
Step 3
read_document returns a document's text as Markdown, part after part: the outline first, then about 20,000 characters at a time, each answer ending with next_from to go on. pages ("2-5", "1,4,7-9") and sheet (a workbook sheet by name) jump straight there; a search match's from opens at the match.
Step 4
get_media on a document returns its preview picture when the file carries one, its format, pages, author and dates, its outline, the start of its text and a 24-hour link to the file — one that works in a folder shared with you too.
Step 5
A document still being read is read on the spot. One that cannot be read says why in plain words — it needs a password, it is a format variant the reader does not know (Word 6/95, .xlsb, .vsd), it is too large, damaged or empty — and stays in the Brain as a file. Too large means over 32 MB for a PDF or Illustrator file, 50 MB for an Office 97–2003 file or Outlook message and 60 MB for the rest; a Photoshop file is read from its first 4 MB and a text file from its first 16 MB, whatever their size. A scanned PDF has no words to read: no OCR is done.
Reference
Generated from the server's definitions, so it is exactly what a client sees. Required parameters are marked *; a default means the parameter can be left out.
The balance, the plan, and every output the user has made.
Get credit balance
Return the signed-in Lumen Brain user's credit balance, plan, the date the next monthly credits arrive (a live subscription only — the Free plan adds none monthly) and Brain storage (bytes used against the room the plan gives, plus any storage an admin gifted on top). The balance comes in two parts: the plan's credits — spent first, carried over to the next month up to one month's allowance, expiring when the plan ends — and credits that never expire (top-ups, gifts, welcome credits). Free to call.
List generations
List the signed-in user's recent Lumen Brain generations with model, credits spent, a 24-hour signed download URL and, while the output's link is open (it is by default), its permanent public URL — the one to paste into other tools or docs. Every item carries its generation id: pass ids to save_output_to_brain to file outputs in a Brain folder, or as reference_generation_ids / reference_video_generation_id to render from them again. Renders still on their way are named first with their job ids (wait_for_generations returns them once done), so the newest finished item is never taken for one still rendering. search finds an earlier output by words of its prompt. Free to call.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| kind | "image" | "edit" | "video" | "tts" | "stt" | "design" | Only return generations of this kind (design = a Design Studio export). | |
| limit | integer (1–50) | 10 | How many to return. |
| search | string (≤ 200 chars) | Words from the prompt, to find an earlier output — e.g. "red shirt". |
Pictures and clips from a prompt, references and presets, rendered in the background; waiting for them when a next step needs them; batches; uploads from the chat.
Generate image
Start an image in Lumen Brain and return as soon as the job is running — the render goes on in the background, so the conversation is free again at once. The inline preview follows the job and shows the picture the moment it lands (usually well within a minute); it is saved to the gallery either way, and a failed render refunds its credits. When a next step in the same task needs the finished picture — placing it in a design, animating it, filing it, reviewing it — call wait_for_generations with the returned job_id: it returns the generation id and links; otherwise do not wait. Models and credit costs: lite (Nano Banana 2 Lite, 1), fast (Nano Banana 2, 2), nano (Nano Banana, 2), pro (Nano Banana Pro, 4), gptImage2 (GPT Image 2, 6), flare (GPT Image 2.6 Flare, 6 — OpenAI's fast default, about twice the speed of GPT Image 2), sunburst (GPT Image 2.6 Sunburst, 8 — OpenAI's most precise, for detailed edits and finished campaign work), qwen (Qwen Image, 2 — Alibaba's open model, quick renders and edits; 1K only), qwen3 (Qwen Image 3, 2 — Alibaba's newest, the sharpest text and layouts, 1K or 2K; hosted by Alibaba, so its content rules apply), seedream (Seedream 5.0 Pro, 2 — ByteDance's newest: it thinks a long brief through, holds a dense layout together and writes words into the picture in fourteen languages, 1K or 2K), seedreamLite (Seedream 5.0 Lite, 1 — the same family, quicker and a third of the price), soul2 (SOUL 2, 1 — Higgsfield's photographic people model), grok (Grok Image 2.0 by xAI, 2 — renders and edits), recraft (Recraft V4.1 Pro, 5 — typography and design, 1K or 2K), and ideogram and ideogramHigh (Ideogram 4.5, 2 and 6 — posters, logos and lettering, 1K or 2K, at its medium and high quality; it renders, and edits the first reference picture with up to four more as references), and krea and kreaLarge (Krea 2 Medium and Large, 1 and 2 — Krea's own model: Medium for illustration, anime and painterly looks, Large for photoreal with raw grain and texture; 1K only. Its strength is style transfer: reference pictures set the style rather than being edited, each at its own strength — style_strength or style_strengths, -2 to 2, default 0.6, below 0 steering away from that look — and creativity sets how freely the prompt is read); 2K output costs 1.75x and 4K 3x the base credits, rounded up. Reference images (inline base64 and/or media-library assets) drive image-to-image work — edits, restyles, or keeping a face or product consistent — up to the model's own cap: lite, nano, qwen and qwen3 take 3, grok 4, ideogram and ideogramHigh 5 (the first is the picture edited), fast 6, the Seedream pair 10, krea and kreaLarge 10 (as style references — nothing is edited), pro 14, gptImage2, flare and sunburst 16; soul2 and recraft take none. Large Brain pictures are sent as a 2048 px copy so the references fit one job. An earlier output goes in the same way — "with this one, now…" — as reference_generation_ids (or reference_latest_output for the newest), read straight from the gallery. For SEVERAL images (a batch, variations, N of something) call queue_generations instead of repeating this tool. For a styled look, pass a curated preset name (see list_presets) as preset — the platform applies the style server-side at no extra credit cost. If the prompt refers to a picture the user attached to the chat and no reference reached the call, nothing is rendered: the tool opens the upload widget in the chat instead (no credits spent) — tell the user to drop the file in, then call get_upload_session and run again with reference_asset_ids.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| prompt * | string | What to generate. | |
| model | "lite" | "fast" | "nano" | "pro" | "gptImage2" | "flare" | "sunburst" | "qwen" | "soul2" | "grok" | "recraft" | "qwen3" | "seedream" | "seedreamLite" | "ideogram" | "ideogramHigh" | "krea" | "kreaLarge" | "fast" | Image model key: lite | fast | nano | pro | gptImage2 | flare | sunburst | qwen | qwen3 | seedream | seedreamLite | soul2 | grok | recraft | ideogram | ideogramHigh | krea | kreaLarge. |
| aspect | "1:1" | "16:9" | "9:16" | "4:3" | "3:4" | "3:2" | "2:3" | "4:5" | "5:4" | Aspect ratio. Left out, it is the preset's own frame when a preset is passed (list_presets shows each one's), else 1:1. | |
| resolution | "1K" | "2K" | "4K" | "1K" | Output resolution as exposed by the model's API: Gemini flash models and GPT Image 2 render 1K-class output (GPT Image's concrete size is fixed per aspect ratio); Nano Banana Pro supports 1K, 2K or 4K, Qwen Image 3, the Seedream 5.0 pair, Recraft V4.1 Pro and Ideogram 4.5 1K or 2K; Krea 2 1K only. Unsupported values fall back to the model's default. |
| reference_image_base64 | string (≤ 8000000 chars) | Optional reference image to edit or stay consistent with: a data URL or raw base64 image. | |
| reference_image_mime | string | MIME type when reference_image_base64 is raw base64. Defaults to "image/png". | |
| reference_asset_id | id | Optional media-library asset id to use as a reference (from list_media). | |
| reference_asset_ids | array of id (≤ 16 items) | Media-library asset ids when several reference images should go in together (capped at the model's own limit: lite/nano/qwen/qwen3 3, grok 4, ideogram/ideogramHigh 5, fast 6, seedream/seedreamLite 10, krea/kreaLarge 10, pro 14, gptImage2/flare/sunburst 16). | |
| reference_folder | string | Media-library folder path of a reference asset, e.g. "/Brand/Logos" — one of the user's own, or the id of a folder shared with them. | |
| reference_name | string | Media-library asset name inside reference_folder. | |
| reference_generation_ids | array of id (≤ 16 items) | Earlier outputs to use as reference images, by generation id — how "with this one, now…" is rendered from something generated before (every generate result and list_generations item carries the id). Read from the gallery server-side; nothing is re-uploaded. | |
| reference_latest_output | boolean | Use the newest finished image output as a reference — "with this one" when its id is not at hand. | |
| preset | string | Curated preset name from list_presets. The platform rewrites the prompt to the preset's style server-side and attaches the preset's own reference images — pass just the name, nothing else is needed. | |
| brain_folder | string | Brain folder to file the finished image in as it lands, e.g. "/Tim Asher/Black shirt" — created if missing; the image lands in the gallery as well. Use it whenever the user wants the output in a Brain folder. An output rendered earlier is filed with save_output_to_brain. | |
| creativity | "raw" | "low" | "medium" | "high" | Krea 2 only (krea, kreaLarge): how freely the prompt is read — raw renders only what it says, low stays close to it (keeps a style picture's look cleanest), medium is Krea's default, high takes liberties with mood, light and framing. | |
| style_strength | number (-2–2) | Krea 2 only: how strongly every reference picture's style counts, from -2 to 2 (default 0.6). Around 0.3 is a tint, 1 lets that look dominate, and below 0 steers the render away from it. | |
| style_strengths | array of number (≤ 10 items) | Krea 2 only: one strength per reference picture, in the order they are read — the inline picture, then reference_asset_ids followed by reference_asset_id (an id given twice counts once), then the newest output when reference_latest_output is set, then reference_generation_ids, then the named asset. A picture without one takes style_strength, or 0.6. |
Generate video
Queue a short video clip (Gemini Omni Flash 1.2, Gemini Omni Flash 1.1, Gemini Omni Flash, MiniMax Hailuo H3, Kling 4.0, Kling 4.0 Edit, Kling 3.0, Kling 2.6 Pro, Kling 2.5 Turbo Pro, PixVerse 6, Seedance 2.0, Seedance 2.5, Wan 3.0, Genjutsu or Cinema Studio 4.0) in Lumen Brain. Returns at once with a job id: the clip renders in the background, the inline preview shows it the moment it lands, and it is saved to the user's gallery — call list_generations later for its signed URL. Credits are reserved on queueing and refunded automatically if the render fails. Several clips? Call this once per clip, back to back — they render in parallel, each with its own preview — or use queue_generations for a large batch of one prompt. Costs 1-8 credits per second on Gemini Omni Flash 1.1 — the default, and with PixVerse 6 and Wan 3.0 the cheapest model here — 3 on Gemini Omni Flash, 3-4 on Hailuo H3, 5-7 on Kling 4.0, 6-8.4 per second of the clip on Kling 4.0 Edit, 3-8 on Kling 3.0, 3 on Kling 2.6 Pro, 2 on Kling 2.5 Turbo Pro, 1-2 on PixVerse 6, 3-6 on Seedance 2.0, 4-9 on Seedance 2.5, 1-4 on Wan 3.0, 5-10 on Cinema Studio 4.0, and on Genjutsu 3-4 on Hailuo H3 or 1-8 on Gemini Omni, depending on resolution (e.g. 6s = 6-60). GENJUTSU is motion transfer: a clip goes in (as reference_video_url / reference_video_asset_id / reference_video_generation_id — it renders from nothing else) and the same motion, camera and timing come out, recast with whoever or whatever takes over. It renders on one of two engines. Key genjutsu runs on Hailuo H3: a clip of 5-15s and up to 9 cast pictures through reference_asset_ids — the people, the product, the clothes — named @Image1, @Image2 … in the prompt ("@Image1 replaces the dancer"); once a picture is passed the prompt is optional; 768p at 3 or 2K at 4 credits/s. Key genjutsuOmni runs on Gemini Omni, for less: a clip of 3-9s under 32 MB and up to 9 cast pictures the same way — Omni reads pictures and a clip together on its reference endpoint, so the clip is cut on the server into up to three consecutive parts of under 3s; once a picture is passed the prompt is optional, and without pictures the prompt is required and says who or what takes over ("a chrome robot replaces the dancer"); 360p at 1, 720p at 3, 1080p at 4, 4k at 8 credits/s. On either engine the output runs as long as the clip, so its length is read from the clip, and the duration parameter is only used for a clip given by URL, or a Brain or gallery clip whose length is not recorded — pass the clip's own length then. CINEMA STUDIO 4.0 (key cinema4) is Seedance 2.5 with Higgsfield's direction on top — 4-30s, 480p or 720p, up to 30 reference pictures, a reference clip and sound — and takes direction, the Director's Panel as fields: genre, era, pacing, light, camera_model, camera_lens, camera_aperture, camera_movement, color_palette, each optional (left out, the model reads the prompt for it). A reference clip on Cinema Studio is billed too: Higgsfield meters the clip's seconds together with the render's at 0.6× the rate, so the price covers both. KLING 4.0 (key kling4) and KLING 4.0 EDIT (key kling4Edit) are in BETA — administrators only until fal opens their API; anyone else is refused, so do not offer them to other users. Kling 4.0 renders one take of 3-30s with sound at 720p (5 credits/s) or 1080p (7): one picture is the starting frame, while 2-10 pictures (reference_asset_ids) or a reference clip of 3-30s go in as references, named @Image1, @Image2 … and @Video1 in the prompt; a reference clip is billed too, its seconds and the render's together at 0.6× the rate and never less than twice the render's. Kling 4.0 Edit changes a clip of 3-30s, up to 200 MB (reference_video_url / reference_video_asset_id / reference_video_generation_id — it renders from nothing else): the prompt says what to change (the clip is @Video1, and up to 4 pictures ride along as @Image1 …), and the clip comes back as long as it went in, with its framing and its sound kept, at 6 credits per second of the clip at 720p or 8.4 at 1080p. Optionally animate a reference image, either inline base64 or a media-library asset. Seedance 2.0, Seedance 2.5, Hailuo H3 and Gemini Omni Flash 1.1 read a whole SET instead — up to 9 pictures (Wan 3.0 up to 4) through reference_asset_ids (a character, a second character, the place they are in), arriving as @Image1, @Image2 … for the prompt to name ("the character from @Image1, in the room from @Image2"); on Seedance a set caps the output at 720p the way a clip does. On Omni Flash 1.1 one picture alone is the starting frame, while two or more — or any beside a reference clip — go to its reference endpoint as a set. Gemini Omni Flash 1.1, Hailuo H3, both Seedance generations and Wan 3.0 also take a reference VIDEO by public URL (reference_video_url): Omni Flash 1.1 carries the clip on or edits it — that is how a scene grows past 10s, 10s at a time to Google's 40s ceiling — and with pictures beside it reads the clip and the pictures together: its reference endpoint takes the clip's last 9s, cut on the server into three consecutive parts under 3s, and renders a new shot of the chosen 3-10s; Hailuo, Seedance and Wan use the clip as a reference to continue, restyle or borrow motion from; a clip caps the output at 720p on Seedance and Wan. Both Seedance generations also take SOUND — up to 3 MP3/WAV files by public URL (reference_audio_urls, 15s and 15 MB each), which arrive as @Audio1, @Audio2 …: say in the prompt what each is for ("BGM references @Audio1", "the voice references @Audio1"). Sound needs a reference image or clip with it, and caps the output at 720p just as a clip does. A clip already in the Brain (uploaded through media_upload_widget, or listed by list_media / search_brain with kind "video") goes in as reference_video_asset_id instead of a URL. For a styled look, pass a curated preset name (see list_presets) as preset — the platform applies the style server-side at no extra credit cost. If the prompt refers to a picture or clip the user attached to the chat and no reference reached the call, nothing is queued: the tool opens the upload widget in the chat instead (no credits reserved) — tell the user to drop the file in, then call get_upload_session and run again with reference_asset_ids or reference_video_asset_id.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| prompt | string | "" | What the clip should show. Required, except on genjutsu once a cast picture is passed: the clip and the pictures are enough there, and a prompt only says who plays whom. |
| model | "omni12" | "omni11" | "hailuo" | "kling4" | "kling4Edit" | "kling3" | "kling26" | "kling25" | "pixverse6" | "seedance" | "seedance25" | "wan3" | "genjutsu" | "genjutsuOmni" | "cinema4" | "omni12" | Video model: "omni12" = Gemini Omni Flash 1.2 via fal.ai, Google's newest (3-15s, 360p at 1 credit/s, 720p at 3, 1080p at 4, 4k at 7 — the default, and with PixVerse 6 and Wan 3.0 the cheapest), "omni11" = Gemini Omni Flash 1.1 via fal.ai (3-10s, 360p at 1 credit/s, 720p at 3, 1080p at 4, 4k at 8), "hailuo" = Hailuo H3 by MiniMax (5-15s, 768p at 3 credits/s, 2K at 4), "kling4" = Kling 4.0 via fal.ai (beta, administrators only; 3-30s with sound, 720p at 5 credits/s, 1080p at 7; up to 10 pictures and a reference clip, which is billed too), "kling4Edit" = Kling 4.0 Edit via fal.ai (beta, administrators only; needs a clip of 3-30s and runs as long as it; up to 4 reference pictures; 720p at 6, 1080p at 8.4 credits per second of the clip), "kling3" = Kling 3.0 via Higgsfield (3-15s, 720p or 1080p at 3, 4k at 8), "kling26" = Kling 2.6 Pro via Higgsfield (3 credits/s, 5 or 10s), "kling25" = Kling 2.5 Turbo Pro via Higgsfield (2 credits/s, 5 or 10s), "pixverse6" = PixVerse 6 via Higgsfield (3-15s, 360p/540p at 1, 720p/1080p at 2), "seedance" = Seedance 2.0 by ByteDance via Higgsfield (4-15s, 480p at 3, 720p at 6), "seedance25" = Seedance 2.5 by ByteDance via Higgsfield (4-30s, 480p at 4, 720p at 9), "wan3" = Wan 3.0 by Alibaba via Higgsfield (4-15s, 480p at 1, 720p at 2, 1080p at 4), "genjutsu" = Genjutsu, motion transfer on Hailuo H3 (needs a reference clip of 5-15s and runs as long as it; up to 9 cast pictures; 768p at 3, 2K at 4), "genjutsuOmni" = Genjutsu on Gemini Omni (needs a reference clip of 3-9s and runs as long as it; up to 9 cast pictures, or a prompt that says who takes over; 360p at 1, 720p at 3, 1080p at 4, 4k at 8), "cinema4" = Cinema Studio 4.0 by Higgsfield (4-30s, 480p at 5, 720p at 10; takes direction). |
| duration | integer (1–30) | Clip length in seconds (default 6) — Omni Flash 1.2 supports 3-15s, Omni Flash 1.1 3-10s, Omni 1-10s, Hailuo 5-15s, Kling 4.0 3-30s, Kling 3.0 3-15s, Kling 2.6 and 2.5 5 or 10s, PixVerse 6 3-15s, Seedance 2.0 4-15s, Seedance 2.5 4-30s, Wan 3.0 4-15s, Cinema Studio 4.0 4-30s; out-of-range values are clamped. Genjutsu and Kling 4.0 Edit run as long as their clip and ignore this for a Brain or gallery clip whose length is known. Genjutsu needs it for a clip given by URL or one whose length is unknown — pass that clip's own length. Kling 4.0 Edit does not: its clip is measured. | |
| aspect | "16:9" | "9:16" | Clip aspect ratio. Left out, it is the preset's own frame when a preset with a 16:9 or 9:16 frame is passed, else "16:9". Genjutsu and Kling 4.0 Edit follow the clip they are given where its shape is known (a Brain or gallery clip), and so do Gemini Omni Flash 1.2 and 1.1 with a clip unless a ratio is passed; for a clip given by URL, pass the clip's own ratio. | |
| direction | object | Cinema Studio 4.0 only: the Director's Panel. Set any of the controls; a control left out is read off the prompt by the model. Pacing runs single-shot, calm, dynamic, chaotic; the camera moves are Higgsfield's own presets (dolly-in, whip-pan, drone-orbit, snorricam …); the colour palettes are their named looks (neon-rain-at-midnight, amber-wasteland …). | |
| resolution | "360p" | "480p" | "540p" | "720p" | "768p" | "1080p" | "2K" | "4k" | Output resolution for Gemini Omni Flash 1.2 (360p at 1 credit/s, 720p at 3, 1080p at 4, 4k at 7) and 1.1 (360p at 1, 720p at 3, 1080p at 4, 4k at 8) — 4k is upscaled from the render, Hailuo H3 (768p at 3, 2K at 4), Genjutsu (768p at 3 or 2K at 4 on genjutsu; 360p at 1, 720p at 3, 1080p at 4, 4k at 8 on genjutsuOmni), Kling 4.0 (720p at 5, 1080p at 7), Kling 4.0 Edit (720p at 6, 1080p at 8.4 per second of the clip), Kling 3.0 (720p or 1080p at 3, 4k at 8), PixVerse 6 (360p/540p at 1, 720p/1080p at 2), Seedance 2.0 (480p at 3, 720p at 6), Seedance 2.5 (480p at 4, 720p at 9), Wan 3.0 (480p at 1, 720p at 2, 1080p at 4) and Cinema Studio 4.0 (480p at 5, 720p at 10). Ignored by Kling 2.6 and Kling 2.5. Left out, a model renders its cheapest tier, except Hailuo H3 and genjutsu (2K), and genjutsuOmni and PixVerse 6 (720p). | |
| reference_image_base64 | string (≤ 8000000 chars) | Optional reference image to animate: a data URL or raw base64 image. | |
| reference_image_mime | string | MIME type when reference_image_base64 is raw base64. Defaults to "image/png". | |
| reference_video_url | url | Optional reference clip as a public https URL (MP4/MOV), for the models that take a video input: Gemini Omni Flash 1.2 and 1.1 (up to 40s in — the render continues or edits it; beside pictures, its last 9s are read with them), Hailuo H3 (2-15s), Kling 4.0 (3-30s, up to 200 MB, as a reference — billed too), Kling 4.0 Edit (3-30s, up to 200 MB — the clip it changes; it renders from nothing else), Seedance 2.0 (2-15s), Seedance 2.5 (2-30s), Wan 3.0 (2-15s), Cinema Studio 4.0 (2-30s) and Genjutsu (5-15s on genjutsu, 3-9s on genjutsuOmni — the clip whose motion is recast; Genjutsu renders from nothing else). | |
| reference_video_asset_id | id | Optional Brain video asset id (kind "video" in list_media / search_brain, or an upload from media_upload_widget) to use as the reference clip — the same models and rules as reference_video_url, which it replaces. | |
| reference_audio_asset_ids | array of id (≤ 3 items) | Brain sound-file ids (kind "audio" in list_media / search_brain, or an upload from media_upload_widget) to use as the render's music or voice track — Seedance 2.0 and Seedance 2.5 only, up to 3. This is how a local file goes in: open media_upload_widget, have the user drop the MP3 or WAV in, then pass the ids here. They arrive as @Audio1, @Audio2 ... and the prompt says what each is for. | |
| reference_audio_urls | array of url (≤ 3 items) | Optional sound files as public https URLs (MP3 or WAV, up to 3, 15s and 15 MB each), for Seedance 2.0 and 2.5 only. They reach the render as @Audio1, @Audio2 … — say in the prompt what each is for ("BGM references @Audio1", "the voice references @Audio1"). A sound file needs a reference image or clip alongside it, and it caps the output at 720p the way a clip does. | |
| reference_asset_id | id | Optional media-library asset id to animate (from list_media). | |
| reference_asset_ids | array of id (≤ 10 items) | Media-library asset ids when SEVERAL reference pictures should go in together — a character, a second character, the place they are in. Kling 4.0 (up to 10), Seedance 2.0, Seedance 2.5, Hailuo H3 and Gemini Omni Flash 1.2 and 1.1 (up to 9), Wan 3.0 and Kling 4.0 Edit (up to 4); they arrive as @Image1, @Image2 … and the prompt names them ("the character from @Image1, in the room from @Image2"). Omni Flash 1.1 also reads them beside a reference clip. Every other model animates a single starting frame, so pass one id here or use reference_asset_id. | |
| reference_folder | string | Media-library folder path of the reference asset, e.g. "/Brand/Logos" — one of the user's own, or the id of a folder shared with them. | |
| reference_name | string | Media-library asset name inside reference_folder. | |
| reference_generation_id | id | An earlier image output as the starting frame, by generation id — "animate this one" (every generate result and list_generations item carries the id). Read from the gallery server-side. | |
| reference_video_generation_id | id | An earlier clip output as the reference clip, by generation id, for the video-to-video models — the same role as reference_video_asset_id. | |
| reference_latest_output | boolean | Use the newest finished output — "with this one" when its id is not at hand: an image becomes the starting frame, a clip the reference clip. | |
| preset | string | Curated preset name from list_presets. The platform rewrites the prompt to the preset's style server-side; when no reference image is supplied, the preset's own reference becomes the starting frame. | |
| brain_folder | string | Brain folder to file the finished clip in, e.g. "/Tim Asher/Black shirt" — created if missing; the clip lands in the gallery as well. Use it whenever the user wants the output in a Brain folder. |
Wait for generations
Wait for renders that generate_image, generate_video or queue_generations started, and get their outputs: for each finished job its generation id (what create_design, update_design, save_output_to_brain and the generate tools' reference_generation_ids take), a 24-hour signed URL and, while the output's link is open, its permanent public link. The generate tools return as soon as a job is running, so the conversation stays free — call this only when a next step in the same task needs the finished output (placing a picture in a design, animating it, filing it, reviewing it). Waits up to timeout_seconds (default 45) for all of them; any still rendering by then is reported as such — call again to keep waiting. A failed render is reported with its reason; its credits are refunded. Free to call.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| job_ids * | array of id (≤ 25 items) | The jobs to wait for: job_id from generate_image or generate_video, job_ids from queue_generations. | |
| timeout_seconds | integer (5–55) | 45 | How long to wait at most, in seconds. |
Queue generations
Queue one or many generations to render in the background instead of waiting for each one. ALWAYS use this instead of repeated generate_image / generate_video calls when the user asks for several images or clips (a batch, variations, N of something): every job is enqueued at once, credits are reserved up-front and refunded on failure, the studios show live progress tiles, and results land in the gallery — check list_generations a few minutes later. Same models, credit costs, reference-image rules and preset support as the generate tools; video jobs also take a Brain clip as reference_video_asset_id for the video-to-video models. Video jobs on Seedance 2.0, Seedance 2.5, Hailuo H3 and Gemini Omni Flash 1.1 read up to 9 reference pictures at once (Kling 4.0 up to 10, Wan 3.0 and Kling 4.0 Edit up to 4) (a character, a second character, the place they are in), arriving as @Image1, @Image2 … for the prompt to name; every other video model animates the first as its starting frame, and so does Omni Flash 1.1 given one picture and no clip. Earlier outputs go in too: reference_generation_ids (images), reference_video_generation_id (a clip, video jobs) or reference_latest_output for the newest one — read from the gallery server-side, nothing re-uploaded. Returns the queued job ids immediately. If the prompt refers to a picture or clip the user attached to the chat and no reference reached the call, nothing is queued: the tool opens the upload widget in the chat instead — tell the user to drop the file in, then call get_upload_session and run again with reference_asset_ids (or reference_video_asset_id).
| Parameter | Type | Default | What it is |
|---|---|---|---|
| kind | "image" | "video" | "image" | What to render. |
| prompt | string | "" | What every queued job should generate. Required, except for video jobs on genjutsu that pass a cast picture, where the clip and the pictures are enough. |
| model | string | "fast" | Model key. Images: lite | fast | nano | pro | gptImage2 | flare (GPT Image 2.6 Flare) | sunburst (GPT Image 2.6 Sunburst) | qwen (Qwen Image, Alibaba's open model) | qwen3 (Qwen Image 3) | seedream (Seedream 5.0 Pro, dense layouts and words in the picture) | seedreamLite (Seedream 5.0 Lite, quicker and cheaper) | soul2 (SOUL 2, Higgsfield's photographic people model, 1 credit) | grok (Grok Image 2.0 by xAI, renders and edits) | recraft (Recraft V4.1 Pro, typography and 2K design) | ideogram (Ideogram 4.5, posters, logos and lettering; renders, and edits the first reference picture) | ideogramHigh (Ideogram 4.5 at its highest quality) | krea (Krea 2 Medium, illustration, anime and painterly looks; reference pictures set the style, each at its own strength — see style_strength and creativity) | kreaLarge (Krea 2 Large, photoreal with raw grain and texture). Videos: omni12 (Gemini Omni Flash 1.2, Google's newest and the default: 3-15s, 4k at 7 credits/s) | omni11 (Gemini Omni Flash 1.1: 3-10s) | hailuo | kling4 (Kling 4.0 — beta, administrators only: 3-30s with sound, up to 10 pictures and a reference clip of 3-30s, which is billed too) | kling4Edit (Kling 4.0 Edit — beta, administrators only: changes a reference clip of 3-30s the way the prompt says and runs as long as it, sound kept; up to 4 pictures) | kling3 | kling26 | kling25 | pixverse6 | seedance | seedance25 | wan3 | genjutsu (Genjutsu, motion transfer on Hailuo H3 — needs a reference clip of 5-15s, runs as long as it, takes up to 9 cast pictures, and the prompt as optional once one is passed) | genjutsuOmni (Genjutsu on Gemini Omni, for less — a reference clip of 3-9s, up to 9 cast pictures, and the prompt as optional once one is passed) | cinema4 (Cinema Studio 4.0 — Seedance 2.5 under Higgsfield's direction; takes direction, and bills a reference clip's seconds too). |
| count | integer (1–25) | 1 | How many jobs to queue with this prompt (credits are reserved for each). |
| aspect | string | Aspect ratio — images default to the preset's own frame when a preset is passed, else "1:1"; videos to a preset's 16:9 or 9:16 frame, else "16:9" (or "9:16"). Genjutsu follows its clip's own shape where it is known, and so do Gemini Omni Flash 1.2 and 1.1 with a clip unless a ratio is passed. | |
| resolution | string | Output resolution: images 1K/2K/4K where the model supports it; videos 480p/720p/1080p for the fal models (720p or 1080p on Kling 4.0 and Kling 4.0 Edit). | |
| duration | integer (1–30) | Video only: clip length in seconds (clamped to the model's range). Genjutsu and Kling 4.0 Edit run as long as their reference clip and ignore this where the clip's length is known. Genjutsu needs it for a clip whose length is unknown — pass the clip's own length; Kling 4.0 Edit measures its clip. | |
| direction | object | Cinema Studio 4.0 only: the Director's Panel — genre, era, pacing, light, camera_model, camera_lens, camera_aperture, camera_movement, color_palette; a control left out is read off the prompt. | |
| creativity | "raw" | "low" | "medium" | "high" | Krea 2 image jobs only (krea, kreaLarge): how freely the prompt is read — raw renders only what it says, low stays close (keeps a style picture's look cleanest), medium is Krea's default, high takes liberties with mood, light and framing. | |
| style_strength | number (-2–2) | Krea 2 image jobs only: how strongly every reference picture's style counts, -2 to 2 (default 0.6); around 0.3 is a tint, 1 lets that look dominate, below 0 steers away from it. | |
| style_strengths | array of number (≤ 10 items) | Krea 2 image jobs only: one strength per reference picture, in the order they are read — the inline picture, then reference_asset_ids, then the newest output when reference_latest_output is set, then reference_generation_ids. A picture without one takes style_strength, or 0.6. | |
| preset | string | Curated preset name from list_presets — applied server-side when each job runs. | |
| reference_asset_ids | array of id (≤ 16 items) | Media-library asset ids to use as reference/input images, up to the model's cap (on a video model that animates one picture, the first becomes the starting frame). | |
| reference_image_base64 | string (≤ 8000000 chars) | One inline reference image: a data URL or raw base64. | |
| reference_image_mime | string | MIME type when reference_image_base64 is raw base64. Defaults to "image/png". | |
| reference_generation_ids | array of id (≤ 16 items) | Earlier outputs to use as reference images, by generation id — "with this one, now…" — up to the model's cap (on a video model that animates one picture, the first becomes the starting frame). Read from the gallery server-side. | |
| reference_latest_output | boolean | Use the newest finished output — "with this one" when its id is not at hand: an image as a reference (the starting frame for a video job), a clip as the reference clip of a video job. | |
| reference_video_asset_id | id | Video only: a Brain video asset id (kind "video" in list_media / search_brain) used as the reference clip — Omni Flash 1.2 and 1.1 continue or edit it (beside pictures, they read the clip's last 9s with them); Hailuo, Seedance, Wan 3.0 and Kling 4.0 use it as a reference; Kling 4.0 Edit changes it. | |
| reference_video_generation_id | id | Video only: an earlier clip output as the reference clip, by generation id — the same role as reference_video_asset_id. | |
| brain_folder | string | Brain folder to file every finished output in, e.g. "/Tim Asher/Black shirt" — created if missing; outputs land in the gallery as well. Use it whenever the user wants the results in a Brain folder. |
List curated presets
List Lumen Brain's built-in curated presets — style packs maintained by the platform (film noir, product studio, anime, analog film, UGC, cyberpunk and more). Each has a name, a title and a short description of the look it produces. To use one, pass its name as the preset parameter of generate_image or generate_video: the platform then rewrites the prompt to match the style and attaches the preset's reference images server-side. A preset with a frame (9:16 for a text-message or notes-app story, 4:5 for a feed card) renders in it when no aspect is passed. The style brief itself is not exposed — only the preset name is needed. Check this before generation work when the user asks for a styled or on-brand look. Free to call.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| search | string | Filter by name, title or description substring. |
Media upload widget
Open an in-chat upload widget so the user can pick files from their device — images (PNG/JPG/WEBP), video clips (MP4/MOV/WEBM), sound files (MP3/WAV) or documents (PDF, Word, Excel, PowerPoint, Outlook, Photoshop, Illustrator, InDesign, EPS, OpenDocument, EPUB, RTF, text, Markdown, CSV, JSON, HTML, XML), up to 1 GB each and 25 MB for sound — and drop them straight into a media library folder. Call it IMMEDIATELY, without asking first, whenever the user's message refers to a picture, clip, document or file they attached to the chat, pasted, or have on their device ("the photo I attached", "this image", "my logo", "the PDF I sent", "csatolt kép") — chat attachments never reach these tools, and asking the user to request an upload is a wasted turn. Make it the only tool call of that turn and tell the user to drop the files in. The widget uploads directly from the chat; a browser fallback link is included. For a file that is already online (a share link, a CDN file, a Lumen output link) call save_media_from_url instead — no widget needed. Returns a session id; after the user uploads (hosts that support it let the widget message you itself), call get_upload_session to see the stored assets — image ids go into generate_image / queue_generations as reference_asset_ids, a clip's id into generate_video as reference_video_asset_id, a sound file's into generate_video as reference_audio_asset_ids, and a document's text is read with read_document (the server reads one of up to 10 MB as it lands and a larger one within a few minutes; search_brain then finds words inside it).
| Parameter | Type | Default | What it is |
|---|---|---|---|
| folder | string | Target folder path, e.g. "/Brand/Logos". Created if missing. Defaults to /Uploads, where every input the user brings is filed. | |
| label | string | Short note shown in the widget, e.g. 'Product shots'. | |
| max_files | integer (1–20) | 10 | Maximum files the user can upload. |
Get upload session
Check an upload session started with media_upload_widget: returns the status and the assets the user uploaded (id, name, kind, folder) so you can reference them in prompts or with get_media — and read a document's text with read_document.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| session_id * | id | Session id returned by media_upload_widget. |
A Drive-style library of folders with notes, holding pictures, clips, sound files and documents. Browse it, search it — inside documents too — read a picture as a reference or a document page by page, file outputs and links into it, and open a folder to anyone with its link.
List media folders
List the signed-in user's media library folders (Drive-style tree with full paths, e.g. /Brand/Logos) with asset counts, folder notes and, for a folder open to anyone with its link, that link (set_folder_link opens and closes one); after them come the folders other people shared with the user — each with its id, who shared it and as what; a shared folder is named by that id in list_media, get_media and read_document, and its note is its owner's. Start here before prompting from a brand folder. Free to call.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| under | string | Only return folders under this path, e.g. "/Brand". Defaults to the whole tree. |
List media assets
List what the user's media library holds — pictures, video clips, sound files and documents — optionally inside one folder path (e.g. /Brand/Logos), including nested subfolders, or of one kind. Returns the folder's own note first — treat it as authoritative context for any task about this folder — plus asset names, kinds, notes and tags (a document also its format and page count), not the bytes. Use get_media to read a picture (or a clip's poster frame and signed URL) and read_document to read a document's text; a clip's id goes into generate_video / queue_generations as reference_video_asset_id, and a sound file's (kind "audio") into generate_video as reference_audio_asset_ids. Free to call.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| folder | string | Folder path of the user's own, e.g. "/Brand/Logos", or the id of a folder shared with them (list_folders lists those). Defaults to root. | |
| recursive | boolean | false | Include assets in nested subfolders. |
| kind | "image" | "video" | "audio" | "document" | Only assets of this kind: pictures, clips, sound files or documents. | |
| search | string | Filter by a name or note substring, or an exact tag. | |
| limit | integer (1–100) | 50 | How many to return. |
Search the Brain
Search the user's whole Brain (media library) in one call — every folder at once — by asset names, notes and exact tags, and inside the text of every document kept there (PDF, Word, Excel, PowerPoint, Outlook, Photoshop and the rest). Pictures, clips, sound files and documents alike; each hit has its folder path, kind and format. A document hit lists up to three places the words appear (page, slide or sheet, with a snippet and from): pass its id and that from to read_document to read from there. In documents, accents are optional and Hungarian and English word forms match ("arvizturo" finds "Árvíztűrő", "jelentésből" finds "jelentés", "reports" finds "report"), and "quoted phrases", or and -word work. Several words match only together — on one page, slide or sheet, or in one stretch of a long text — so search one distinctive word first. A query whose whole words find nothing (a quoted phrase aside) is tried again as the starts of words; when a search still misses, try another form of the word or just its first letters. Follow up with get_media for a picture (a clip's poster frame and signed URL) or list_media to browse around a hit; a clip's id goes into generate_video as reference_video_asset_id. Free to call.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| query * | string (≤ 256 chars) | What to look for: a name or note substring, an exact tag, or words in the documents ("quoted phrase", or, -word) — up to 16 words inside documents. | |
| in | "everything" | "names" | "content" | "everything" | Where to look: names, notes and tags (names), the text inside documents (content), or both (everything). |
| limit | integer (1–50) | 20 | How many matches to return. |
Get media asset
Read one stored asset from the media library by asset id or by folder path + name. An image comes back as base64 image content to use as a visual reference when prompting; a video clip comes back as its poster frame plus a 24-hour signed URL, and its id goes into generate_video / queue_generations as reference_video_asset_id; a sound file comes back as a 24-hour link. A document (PDF, Word, Excel, PowerPoint, Photoshop …) comes back as its preview picture when the file carries one, what it is (format, pages, author, dates), where reading it stands, its outline, the first part of its text and a 24-hour link to the file — call read_document for the whole text, page by page. Free to call.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| id | id | Asset id from list_media. | |
| folder | string | Folder path of the user's own, e.g. "/Brand/Logos", or the id of a folder shared with them (list_folders lists those). | |
| name | string | Asset name inside that folder. |
Read a document
Read the text of a document kept in the user's Brain — PDF, Word, Excel, PowerPoint, Outlook and email messages, Visio, Photoshop and Illustrator files, InDesign IDML, XD, EPS, OpenDocument, EPUB, RTF, CSV, text, Markdown, JSON, HTML or XML — page by page (slide by slide, sheet by sheet), as Markdown with headings, lists and tables, plus footnotes, comments and speaker notes. Find it by id (list_media, search_brain) or by folder path + name. Answers are paged at about max_chars characters: the first one opens with the outline, and each ends with next_from — pass it as from to go on; pages ("3", "2-5", "1,4,7-9" — page, slide or sheet numbers) or sheet (a workbook sheet by name) jump straight there, and a search_brain match's from opens at the match. A document still being read is read on the spot; one that cannot be read says why. Use this for a document's words; get_media for a picture or a clip, or for a document's preview, details and a link to the file; search_brain to find which documents mention something. Free to call.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| id | id | The document's asset id (list_media, search_brain). | |
| folder | string | Folder path of the user's own when finding the document by name, e.g. "/Reports", or the id of a folder shared with them (list_folders lists those). Defaults to root. | |
| name | string | The document's name inside that folder, with or without its extension ("Q3 report"). | |
| pages | string (≤ 200 chars) | Only these pages, slides or sheets, by number: "3", "2-5" or "1,4,7-9" (up to 200 at a time). | |
| sheet | string (≤ 200 chars) | Only this sheet of a workbook, by its name (any case). | |
| from | integer (1–1000000) | Where to go on from: the previous answer's next_from, or a search_brain match's from. Pass pages or sheet again to stay inside them. | |
| max_chars | integer (1000–60000) | 20000 | About how many characters of text one answer carries. |
Create media folder
Create a folder (and any missing parents) in the media library, e.g. "/Brand/Logos". Optionally attach a note describing how the folder should be used when prompting.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| path * | string | Full folder path, e.g. "/Brand/Logos". | |
| description | string | Note about this folder, e.g. brand rules or intended use. |
Save media asset
Save a file sent as base64 into the user's media library, in a folder path of their own (created if missing): a picture (PNG, JPEG, WebP, GIF or AVIF), kept as it is, or a document — PDF, Word, Excel, PowerPoint, Outlook, Photoshop, Illustrator, InDesign, EPS, OpenDocument, EPUB, RTF, text, Markdown, CSV, JSON, HTML or XML, told by its name, its type or its bytes — up to 20 MB. A document is stored as a file and read at once, so read_document gives its text page by page and search_brain finds words inside it. Use it to build brand folders that later prompts and bulk runs can read. A video clip reaches the Brain through media_upload_widget. For a file that already has a public URL, call save_media_from_url instead and skip the base64.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| name * | string | Asset name, e.g. 'logo-primary' — or a file name with its extension, e.g. 'brief.pdf', which also says what a document is. | |
| data_base64 * | string | Raw base64 of the file (a data: URL prefix is fine too). | |
| folder | string | Folder path, e.g. "/Brand/Logos". Defaults to root. | |
| mime_type | string | "image/png" | The file's MIME type, e.g. image/png or application/pdf. |
| description | string | Note on how to use this asset in prompts. | |
| tags | array of string | [] | Tags for filtering. |
Save media from a public link
Import a picture, a video clip or a document that already lives at a public http(s) link — a share link, a CDN file, an image on a website, a PDF or an Office file online, or a Lumen Brain output link (lumenbrain.app/o/…) — straight into the user's media library, in a folder path (created if missing; defaults to /Uploads). The server fetches the file, so no base64 travels through the chat: prefer this over save_media whenever the file has a URL, and over media_upload_widget whenever the file is online rather than on the user's device. Images (PNG/JPG/WEBP/GIF/AVIF) and clips (MP4/MOV/WEBM), each up to 32 MB from a link (a bigger file goes in through media_upload_widget); a big image keeps its original in storage with a working copy inline. Documents — PDF, Word, Excel, PowerPoint, Outlook, Photoshop, Illustrator, InDesign, EPS, OpenDocument, EPUB, RTF, text, Markdown, CSV, JSON or XML, up to 32 MB from a link — are told by their bytes, stored as files and read at once when the reader holds them (a PDF or Illustrator file up to 32 MB, an Office 97–2003 file or Outlook message up to 50 MB, other files up to 60 MB, Photoshop and text files from their start whatever their size): the answer says whether the text is ready and how long it is, read_document gives it part by part, and search_brain finds words inside it. Up to 10 links per call; each becomes its own asset, and the returned ids go into generate_image / queue_generations as reference_asset_ids or into generate_video as reference_video_asset_id. Filing spends no credits.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| url | url | One public http(s) link to the file. | |
| urls | array of url (≤ 10 items) | Several links at once (up to 10); each becomes its own asset. | |
| name | string (≤ 200 chars) | Asset name for a single url, e.g. 'packshot-01.png'. Defaults to the file name in the link. | |
| folder | string | Folder path, e.g. "/Brand/Products". Created if missing. Defaults to /Uploads, where every input the user brings is filed. | |
| description | string | Note on how to use the asset(s) in prompts. | |
| tags | array of string | [] | Tags for filtering, e.g. ['brand']. |
Save output to Brain
File finished generations into a Brain folder (created if missing) — the way to put outputs where the user wants them, done server-side, with nothing downloaded or re-uploaded. Pass generation ids (a generate tool's result carries generation_id; list_generations lists them) or latest for the newest N finished outputs. An image becomes a Brain image in the folder (a big one keeps its original in storage with a working copy inline); a clip becomes a Brain clip, ready as a video input. Filing spends no credits and leaves the gallery as it is. To file an output the moment it renders, pass brain_folder to generate_image, generate_video or queue_generations instead; to rearrange later, use move_media.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| folder * | string | Brain folder path, e.g. "/Tim Asher/Black shirt". Created if missing. | |
| generation_ids | array of id (≤ 50 items) | Outputs to file, by generation id. | |
| latest | integer (1–50) | Instead of ids: the newest N finished outputs, most recent first. | |
| kind | "image" | "video" | With latest: only outputs of this kind. | |
| name | string (≤ 120 chars) | Asset name when filing one output. Several are named after their prompts. | |
| description | string (≤ 500 chars) | Note on how to use the asset in prompts. Defaults to the model and the prompt. | |
| tags | array of string (≤ 20 items) | Tags. Defaults to ["generated"]. |
Move media asset
Move a Brain asset (a picture, a clip, a sound file or a document) into another folder, optionally renaming it — for arranging the library: sorting outputs, uploads and documents into the folders they belong to. The target folder is one of the user's own, created if missing — a file stays among its owner's folders, so one in a folder someone shared with the user is not moved out of it. The asset keeps its id, tags and links, and a document keeps its text and search. Find the asset by id (list_media, search_brain, a filing result) or by its current folder path and name.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| id | id | Asset id. | |
| folder | string | Current folder path when finding the asset by name, e.g. "/Uploads". | |
| name | string | Asset name inside folder, when no id is given. | |
| to_folder * | string | Destination folder path, e.g. "/Tim Asher/Black shirt". Created if missing; "/" is the root. | |
| new_name | string (≤ 120 chars) | A new name for the asset. |
Set folder note
Write or update a folder's note in the media library. The note is the folder's persistent context — usage rules, brand guidance, learnings from past work — read back by list_folders and list_media in every future session, so record here anything worth remembering about the folder and its contents. mode=replace rewrites the whole note (an empty string clears it); mode=append keeps the existing note and adds this text on a new line. Only the folder's owner can write. Free to call.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| folder * | string | Folder path, e.g. "/Brand/Logos". | |
| note * | string (≤ 4000 chars) | The note text. With mode=replace an empty string clears the note. | |
| mode | "replace" | "append" | "replace" | replace = overwrite the note; append = add this on a new line after the note. |
Make a folder public, or private again
Opens one of the user's Brain folders to anyone with its link — the folder and everything beneath it, as the Share dialog's general access does — or closes it again. access: viewer (anyone with the link can see and download the files), editor (with a free Lumen Brain account they can also add and change files) or restricted (the link stops working, and people who joined through it lose access; people invited by name keep theirs). The link stays the same across viewer and editor and when the folder is opened again later; new_link: true replaces it, so the old one stops working for everyone who has it. Returns the folder's public page (lumenbrain.app/s/…); while the folder is open, list_media and get_media give each file in it a public link of its own. A folder beneath one that is open stays reachable through that folder's link until that one is closed too — the answer says so. This puts the files on the open internet: confirm the folder and the access with the user first. Only the folder's owner can change its link. Free to call.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| folder * | string | Folder path, e.g. "/Brand/Logos". | |
| access | "viewer" | "editor" | "restricted" | "viewer" | viewer: anyone with the link can see and download; editor: they can also add and change files; restricted: private again. |
| new_link | boolean | false | Replace the link: the old one stops working for everyone who has it. |
Reusable instruction packs the user keeps: brand rules, workflows, styles.
List custom skills
List the user's stored custom skills — reusable instruction packs with a name, title and a "when to use" description — including skills other people shared with them, marked with who shared each. Check this before generation or brand work: when a skill's description matches the task at hand, load it with use_skill and follow it. Free to call.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| search | string | Filter by name, title or description substring. |
Use a custom skill
Load one stored skill's full instructions and activate them for the current task. Once loaded, the skill governs how you work in this session: follow it when writing generation prompts, when choosing models and settings on generate_image/generate_video, and when persisting context (set_folder_note, asset notes). Free to call.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| name * | string | Skill name from list_skills. |
Save a custom skill
Store or update a custom skill so any future session can load it with use_skill. Use this when the user asks to save a workflow, style or set of rules as a skill. Write the description as a trigger ("when to use this") and the instructions as directives covering prompt style, model/settings choices and what context to persist. Saving under an existing name overwrites that skill.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| name * | string | Slug-like skill name, e.g. "brand-ugc-video". Overwrites an existing skill of the same name. | |
| title * | string | Short human title, e.g. "Brand UGC video". | |
| description | string | "" | One or two sentences on when to load this skill, e.g. "Use for every UGC-style ad video." |
| instructions * | string | The skill body: concrete directives for prompts, model choice and context writing. No length limit — full style guides fit. |
The presets everyone has and the user's own designs; reading them layer by layer, filling their slots, editing them with the editor's own moves, checking the layout.
List design templates
The Design Studio's templates: the presets everyone has (the built-in ones referenced as starter:<id>, the ones an admin shares by id) and the user's own saved templates (by id), each with its page size (or canvas: "infinite" for an infinite canvas, whose width × height is only its nominal frame), tags, slots — the bound layers a template fills in: text slots (headline, body, cta…) with the most characters each takes (max_chars) and image slots (image, portrait, logo…) — and its own instructions, the template's system prompt for whoever writes into it: follow it when writing values for create_design. Pass a reference to create_design (with a value per slot) or generate_design_variants; get_design shows a template's slots again with include_doc for its full document. Free to call.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| search | string | Words to match against titles and tags. |
List designs
The user's saved designs and templates in the Design Studio, newest first: id, title, kind, page size — or canvas: "infinite" for an infinite canvas, whose width × height is only its nominal frame — and tags. get_design shows one's slots and document; render_design turns one into a picture. Free to call.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| kind | "design" | "template" | Only designs, or only templates. | |
| search | string | Words to match against titles. | |
| limit | integer (1–100) | 30 |
Get a design
One design or template by reference (a design id, or a starter as starter:<id>): its page size — or canvas: "infinite" for an infinite canvas, which has no page edges and whose width × height is only its nominal frame — its tags and slots; with include_layers a table of its layers — id, name, type, box, what each holds (a shape's kind and settings, a drawing's tool and ink, a connector's ends, route and heads) — the ids edit_design takes; with include_doc the complete design document, the JSON the studio edits, to read or to change and pass back to create_design / update_design as doc (freehand drawings make it long: every stroke is a layer of points). Free to call.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| design * | string | A design id, or starter:<id>. | |
| include_layers | boolean | false | Also return the layers as a table: ids, names, boxes, text, pictures, shapes, drawings, connectors. |
| include_doc | boolean | false | Also return the full document. |
Create a design
Saves a new design (or template) in the user's Design Studio: from a template reference with a value per slot — text as it is; an image slot takes a Brain asset id, a gallery generation id (an output of generate_image / queue_generations), an https link, or "empty" — or from a complete design document you composed (doc, in the shape get_design returns with include_doc) — a page, or an infinite canvas when it carries "canvas": "infinite", whose width × height is then only its nominal frame. Returns the saved design with its slots and an editor link; with render: true also queues a picture of it (see render_design). Free; the AI tools are the ones that spend credits.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| from | string | The template or design to start from: starter:<id> or a design id. Omit when passing doc. | |
| values | object | A value per slot key of the template (see its slots). | |
| doc | any | A complete design document instead of a template. | |
| title | string (≤ 120 chars) | ||
| kind | "design" | "template" | "design" | |
| tags | array of string (≤ 12 items) | ||
| pictures | "web" | "none" | "none" | web: every empty frame gets a fitting picture found on the web for its hint — Gemini (Google Search) or Claude searches the open web for a page showing a public picture and the server files the picture from it (1 credit a picture; a stock library with a key, Wikimedia Commons or Openverse answer for free) — filed in the Brain with its credit; none: frames stay empty. |
| render | boolean | false | Also queue a picture of the design. |
| format | "png" | "jpeg" | "webp" | "pdf" | "png" | |
| scale | number (1–3) | 1 | Pixels per page pixel of the picture; 2 for a sharper copy. |
Update a design
Changes one of the user's designs or templates: pour new values into its slots (text, or a picture as a Brain asset id / gallery generation id / https link / "empty"), replace its document (a page, or an infinite canvas when it carries "canvas": "infinite"), rename it, make it a template or a design again, retag it. Nothing else changes. With render: true also queues a picture of the result.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| design * | string | The design id (a starter cannot be changed — create_design from it instead). | |
| values | object | ||
| doc | any | A complete document to replace the current one. | |
| title | string (≤ 120 chars) | ||
| kind | "design" | "template" | ||
| tags | array of string (≤ 12 items) | ||
| render | boolean | false | |
| format | "png" | "jpeg" | "webp" | "pdf" | "png" | |
| scale | number (1–3) | 1 |
Edit a design's layers
The editor's own moves on one of the user's designs, as a list of operations applied in order: add_text, add_picture, add_frame; add_shape (any of 27 kinds — rect, ellipse, triangle, star or burst with 3–24 points, line, polygons, block arrows, heart, cloud, moon, lightning, speech-bubble, cylinder… — with a fill or "none", an outline, and the kind's one proportion, adjust); add_connector (a line or arrow joining two layers — it follows them when they move — straight, elbow or curved, with heads and dashes: the links of a flowchart or a diagram); add_drawing (a freehand pen, marker or highlighter stroke through page points); add_gradient (a dark scrim, or a round shadow to set under an object); set (any property of one layer — words, font, size, colour, box, slot key, max_chars, a frame's picture or hint, a shape's kind, fill, outline, points and adjust, a drawing's tool, ink and width, a connector's ends, anchors, route, heads and dash; a field the layer does not take is refused), move, resize, remove, duplicate, reorder, align; set_background (a colour, a linear or radial gradient, or a picture), remove_background_picture, resize_page, set_canvas (a page into an infinite canvas — no page edges, its width × height only the nominal frame — or back into a page), set_instructions (the template's system prompt), fill_frames (pictures into the empty frames). On an infinite canvas give new layers an x and y near what is there (one without lands in the nominal frame); resize_page, add_gradient, background pictures and aligning a single layer to the page are refused there. Layers are named by id, name or slot key — get_design with include_layers lists them. Every operation is checked against the document's rules and the whole is saved once; the layers come back as a table, with a warning when copy prints over copy or runs off the page. Free; with render: true a picture is queued.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| design * | string | The design id (a starter cannot be edited in place — create_design from it first). | |
| operations * | array of op: add_text | add_picture | add_frame | add_shape | add_connector | add_drawing | add_gradient | set | move | resize | remove | duplicate | reorder | align | set_background | remove_background_picture | resize_page | set_canvas | set_instructions | fill_frames (≤ 40 items) | What to do, in order. | |
| title | string (≤ 120 chars) | A new title, if wanted. | |
| check | boolean | true | Measure the result for copy printing over copy or (on a page) off the page, and say so. |
| render | boolean | false | Also queue a picture of the result. |
| format | "png" | "jpeg" | "webp" | "pdf" | "png" | |
| scale | number (1–3) | 1 |
Check a design's layout
Measures a design the way the Studio does before a post goes out: which text prints over other text, and which layers run off the page (an infinite canvas has no page to run off). With repair: true the layout is put right the way the Studio's own fit does it — air under grown copy closes first, what sits below moves down, the stack slides up when the page is short, and only then is the type set smaller (to 75% of its size; strict: true allows 60%) — and the repaired design is saved when it is one of the user's own. Free.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| design | string | A design id or starter:<id>. Omit when passing doc. | |
| doc | any | A complete design document to check instead. | |
| repair | boolean | false | Also put the layout right. |
| strict | boolean | false | Let the repair shrink type further (to 60%). |
| save | boolean | true | Save the repaired document over the design. |
Delete a design
Deletes one of the user's designs or templates for good. Pictures already rendered from it stay in the gallery.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| design * | string | The design id. |
Remove a picture's background
The subject cut out of a picture and everything else made transparent: a packshot off its studio floor, a person off their room, a logo off its card. BiRefNet's matting keeps hair, fur and soft edges where a plain mask smears them. Point at a picture in the Brain (asset_id), an output in the gallery (generation_id) or a public link (url). The cut-out is filed as a transparent PNG of its own, beside the original in the same folder, and the original is left as it is. The returned asset id goes straight into create_design, update_design or edit_design as a picture; its outline and shadow follow the subject instead of the frame on their own (cutout: false on the picture layer keeps them on the frame). Free, within the day's allowance the user's plan carries, at most two at once; the answer says how many are left. If hair or fur comes back rough, call again with profile: "fine" — it counts as three of the day's cut-outs.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| asset_id | string (≤ 64 chars) | A picture in the Brain. | |
| generation_id | id | An image output in the gallery. | |
| url | url | A public link to a picture. | |
| profile | "standard" | "fine" | "standard" | "standard" is the everyday pass; "fine" runs the matting model for hair and fur, and counts as three of the day's cut-outs. |
| folder | string (≤ 500 chars) | Where to file the cut-out; the original's own folder otherwise. | |
| name | string (≤ 200 chars) | What to call it; the original's name otherwise. |
Compress a picture
A smaller copy of a picture at high quality — for the web, a site, an email or a social post. Point at a picture in the Brain (asset_id), an output in the gallery (generation_id) or a public link (url). The original is re-encoded at its own pixel size (nothing is cropped or scaled) and the copy is filed beside it, in the same folder with "compressed" after its name and the tag "compressed" (outputs and links go to /Compressed); the original is left as it is. quality: "high" (the default, quality 85, hard to tell from the original), "balanced" (75) or "small" (60). format: "auto" (the default — a JPEG stays a JPEG, anything else becomes WebP, which keeps transparency), "webp", or "original" to keep the file type. A copy that would not be at least 5% smaller is not filed, and the answer says so. PNG, JPEG, WebP and AVIF pictures up to 25 MB and 50 megapixels; free, no credits.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| asset_id | string (≤ 64 chars) | A picture in the Brain. | |
| generation_id | id | An image output in the gallery. | |
| url | url | A public link to a picture. | |
| quality | "high" | "balanced" | "small" | "high" | "high" (85, the default), "balanced" (75) or "small" (60). |
| format | "auto" | "webp" | "original" | "auto" | "auto" keeps a JPEG a JPEG and makes anything else WebP; "webp" always; "original" keeps the file type. |
| folder | string (≤ 500 chars) | Where to file the copy; the original's own folder otherwise. | |
| name | string (≤ 200 chars) | What to call it ("compressed" is added); the original's name otherwise. |
Share a design with everyone (admin)
For an administrator: makes one of their designs or templates a preset everyone has — it appears in every user's Design Studio and in list_design_templates, usable by all and changed by its owner alone — or takes it back. A shared template that started from a built-in starter stands in for that starter. Anyone else is refused.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| design * | string | The design id (one of the admin's own). | |
| for_everyone | boolean | true | false takes a shared design back. |
List design page sizes
The page sizes the Design Studio offers, by group (Social, Ads, Print, Other): each with its id — what generate_design, create_design's documents and edit_design's resize_page and set_canvas take as preset — its name, its pixels and its ratio. Any width × height in pixels is accepted too; these are the formats people actually post, print or pin. Beside the sizes there is the infinite canvas: a board with no page edges for diagrams, flowcharts, mind maps and drawings, whose width × height is only its nominal frame — edit_design's set_canvas turns a design into one (or back into a page), and a document with "canvas": "infinite" is one. Free, needs no sign-in.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| group | "Social" | "Ads" | "Print" | "Other" | Only one group. |
List design fonts
The typefaces a design can name: the built-in families (Google Fonts, by category — sans, serif, display, handwriting — with the weights each ships and whether it has an italic) and the fonts the user uploaded themselves in the Studio, by family name. A layer's font in edit_design and a document's fontFamily take any family listed here; brand fonts passed to generate_design should be from here too. Free to call.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| search | string | Words to match against family names. | |
| category | "sans" | "serif" | "display" | "handwriting" | "mono" | Only built-in families of one category. |
Designs written from a brief, revised by instruction, multiplied from a template; research off the web, pictures off the web, and Claude's eye on the renders.
Generate a design with AI
The AI Post Editor: from a brief — the message, the audience, the tone, the words to use — a complete design is written and saved (layers of copy, colour and picture slots; every piece of copy and every image slot bound as a slot, so it works as a template). Size by preset (ig-square, ig-portrait, ig-story, fb-post, linkedin-post, x-post, yt-thumb, pinterest, fb-cover, linkedin-banner, ad-square, ad-landscape, ad-leaderboard, ad-mrec, a4-portrait, a4-landscape, a5-flyer, business-card, presentation, og-image, email-header, wallpaper) or width × height in px — always a page (an infinite canvas is made with edit_design's set_canvas). Optionally restyle an existing template or design for the brief (from), and pass brand colours, fonts and voice. Models: sonnet (Claude Sonnet 5, 2 credits) or opus (Claude Opus 5, 5 credits). Returns the saved design with its slots; picture slots come back empty — fill them with update_design (a Brain asset or a generate_image output) and render_design for a picture, or pass render: true here to queue one at once. conversation: true also opens the brief as a conversation on the AI Post Editor's Create tab, where the user (or continue_design_conversation) takes it further turn by turn.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| brief * | string (≤ 4000 chars) | What the design is for, in the language the copy should be in. | |
| preset | "ig-square" | "ig-portrait" | "ig-story" | "fb-post" | "linkedin-post" | "x-post" | "yt-thumb" | "pinterest" | "fb-cover" | "linkedin-banner" | "ad-square" | "ad-landscape" | "ad-leaderboard" | "ad-mrec" | "a4-portrait" | "a4-landscape" | "a5-flyer" | "business-card" | "presentation" | "og-image" | "email-header" | "wallpaper" | A page size preset. | |
| width | integer (64–6000) | ||
| height | integer (64–6000) | ||
| model | "sonnet" | "opus" | "sonnet" | |
| from | string | A template or design to restyle and refill: starter:<id> or a design id. | |
| brand | object | ||
| title | string (≤ 120 chars) | ||
| kind | "design" | "template" | "design" | |
| tags | array of string (≤ 12 items) | ||
| pictures | "web" | "none" | "web" | web: every empty frame gets a fitting picture found on the web for its hint — Gemini (Google Search) or Claude searches the open web for a page showing a public picture and the server files the picture from it (1 credit a picture; a stock library with a key, Wikimedia Commons or Openverse answer for free) — filed in the Brain with its credit; none: frames stay empty. |
| research | boolean | false | Search the web on the brief first and write the copy from what is happening now — real names, numbers and dates. 1 credit more, given back when nothing current is found. |
| render | boolean | false | Also queue a picture of the result. |
| format | "png" | "jpeg" | "webp" | "pdf" | "png" | |
| scale | number (1–3) | 1 | |
| conversation | boolean | false | Also keep the brief and the result as a conversation on the AI Post Editor. |
Revise a design with AI
One instruction applied to one of the user's designs by the AI editor — "make the headline shorter and bolder", "swap to a dark palette", "add a price tag bottom right" — and the whole document comes back revised and saved in place, ids and slots kept — on an infinite canvas too, where it lays new nodes out near what is there and joins them with connectors; freehand drawings are kept as they are. sonnet costs 2 credits, opus 5. With render: true also queues a picture of the result.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| design * | string | The design id. | |
| instruction * | string (≤ 2000 chars) | ||
| model | "sonnet" | "opus" | "sonnet" | |
| render | boolean | false | |
| format | "png" | "jpeg" | "webp" | "pdf" | "png" | |
| scale | number (1–3) | 1 |
Generate variants of a template with AI
Bulk content from one template: the AI writes the copy for N versions — from a brief (each variant a different angle), or one variant per line of rows (a pasted spreadsheet column: a product, a quote, a city…) — and each set of values is poured into a copy of the template and saved as its own design. Picture slots keep the template's picture; for each variant the AI's picture prompt is returned, so pictures can be made with queue_generations and placed with update_design. Costs 1 credit per started ten variants on sonnet, 2 on opus. With render: true a picture of every variant is queued. batch: true also writes the run to the AI Post Editor's Bulk tab as a batch (list_bulk_batches), where the user can review, edit and render the posts; project: true gathers the saved variants into one design project (list_design_projects), a document the Studio opens page under page — both for pages: the variants of an infinite canvas stand on their own.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| template * | string | starter:<id> or a design id with slots. | |
| count | integer (1–50) | 5 | |
| brief | string (≤ 2000 chars) | What the versions are about and how they should differ. | |
| rows | string (≤ 8000 chars) | One line per variant — the raw material of each. | |
| model | "sonnet" | "opus" | "sonnet" | |
| save | boolean | true | Save each variant as a design (off: only the values come back). |
| title | string (≤ 100 chars) | Title prefix for the saved variants; numbered. | |
| tags | array of string (≤ 12 items) | ||
| pictures | "web" | "none" | "none" | web: every empty frame gets a fitting picture found on the web for its hint — Gemini (Google Search) or Claude searches the open web for a page showing a public picture and the server files the picture from it (1 credit a picture; a stock library with a key, Wikimedia Commons or Openverse answer for free) — filed in the Brain with its credit; none: frames stay empty. For variants the AI's picture prompt per slot is what is searched for; every variant's frames are fetched one by one, so a large batch takes a while. |
| research | boolean | false | Search the web on the brief first and write the copy from what is happening now — real names, numbers and dates. One search for the batch whatever its size: 1 credit more, given back when nothing current is found. |
| render | boolean | false | Also queue a picture of every saved variant. |
| format | "png" | "jpeg" | "webp" | "pdf" | "png" | |
| scale | number (1–3) | 1 | |
| batch | boolean | false | Also keep the run as a batch on the AI Post Editor's Bulk tab. |
| project | boolean | false | Also gather the saved variants into one design project, as its pages. |
Research a topic on the web
The web read the way the AI Post Editor reads it before writing about something current: a few focused searches planned from the brief, the answers merged, kept to what is about the subject and recent, and returned as dated facts with their sources — names, numbers, dates, prices, results as they stand today. Use it before writing copy about this week, a launch, a price, a result, an event, or anything else that has a date; pass rows (one line per post of a batch) and the plan covers the batch. 1 credit, taken before the search and given back when the web has nothing new to say. generate_design and generate_design_variants run the same search themselves with research: true.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| brief * | string (≤ 2000 chars) | What the copy will be about, in the language the copy should be in. | |
| rows | string (≤ 8000 chars) | One line per post of a batch, so the searches cover all of them. |
Find pictures on the web
Pictures for a few words, found the way the Design Studio fills a frame: the free stock libraries (Pexels, Unsplash, Pixabay, Openverse, Wikimedia Commons) first, and when they have nothing, an engine (Gemini with Google Search, or Claude) searches the open web for a page showing a public picture; the best matches are fetched and filed into the Brain — under /Design Studio/Web pictures unless a folder is given — with the photographer and the licence as the note, so each comes back as a Brain asset id ready for a picture slot (create_design, update_design, edit_design's fill_frames). Free from a stock library; 1 credit for an engine's search that found something. Name the real subject — the product, the person, the place, the event — not a mood.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| query * | string (≤ 200 chars) | What the picture should show, in English. | |
| aspect | "square" | "portrait" | "landscape" | "square" | The shape of the frame it is for; the nearest pictures rank first. |
| count | integer (1–6) | 1 | How many to file. |
| folder | string | A Brain folder to file them in. |
Review finished designs with AI
The art director's check the Studio makes before a post goes out: Claude looks at finished renders (get_design_render must say done) and says, for each, whether it can go out — naming text printing over text, anything cut off by the page edge, copy that cannot be read where it sits, a badge or headline covering the subject of the photograph, an empty frame or a placeholder left in. Taste and tone are not judged. Up to 8 renders a call; 1 credit per started ten. A post that fails is put right with edit_design, check_design_layout (repair: true) or revise_design, then rendered and reviewed again.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| render_ids * | array of id (≤ 8 items) | Finished renders, from render_design / get_design_render. |
Pictures of designs, painted in a browser and handed back as public links.
Render a design
Turns a design into a picture — or a print-ready PDF — saved to the user's gallery with a public link, in the Design Studio's export choices: png, jpeg, webp or pdf (text, shapes and lines drawn as vectors); a pixel scale or a print size in mm, cm or inches at a resolution (a B1 poster: print { unit: "cm", width: 70.7, dpi: 300 }); a quality; a see-through page. Pass a design or template reference (optionally with values for its slots, so a template renders filled without being saved first) or a complete document. Painting takes a browser: where this chat shows Lumen Brain views the picture is painted right here in the result view; otherwise it is painted the moment the user has Lumen Brain open in a browser (the studio paints waiting renders itself), or by opening the returned page_url once. An infinite canvas is painted as what is on it with a margin, at most 8192 px on its long edge. Call get_design_render with the render_id for the picture's links when it is done. Free.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| design | string | A design id or starter:<id>. Omit when passing doc. | |
| values | object | Values per slot, poured in for this render only. | |
| doc | any | A complete design document to render instead. | |
| title | string (≤ 120 chars) | The picture's name in the gallery. | |
| format | "png" | "jpeg" | "webp" | "pdf" | "png" | png, jpeg, webp — or pdf: a print-ready PDF, text, shapes and lines drawn as vectors, pictures at the resolution asked for. |
| scale | number (0.25–3) | 1 | Pixels per page pixel when no print size is given; 2 for a sharper copy. |
| object | A print size instead of a pixel scale: unit mm, cm or in, the width (or the height — the other follows the design's proportions) and a resolution in dpi (300 when left out). A PNG or JPEG gets that many pixels to the inch and says so in its file; a PDF's page is exactly that size. E.g. { unit: "cm", width: 70.7, dpi: 300 } for a B1 poster. | ||
| quality | integer (1–100) | 1–100 for jpeg, webp and a PDF's pictures. | |
| transparent | boolean | png and webp: leave the page's own colour out. |
Get a design render
Where a render stands, by its render_id: pending (not painted yet — the user's open Lumen Brain or the page_url paints it), failed (with why), or done with the picture's links — a public link that never expires and a signed one — and, when it can be fetched, a small preview as image content. Free to call.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| render_id * | id |
Render every page of a project
A picture — or a print-ready PDF — of every page of one of the user's projects, in order — one render each, saved to the gallery with public links as render_design does, in the same export choices (a print size measures the first page, and every page is printed at its scale). Painting takes a browser: where this chat shows Lumen Brain views the pages are painted right here; otherwise the moment the user has Lumen Brain open, or by opening a page_url once. get_design_render reads each render by id. Free.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| project * | id | ||
| pages | array of id | Only these pages; every page when omitted. | |
| format | "png" | "jpeg" | "webp" | "pdf" | "png" | png, jpeg, webp — or pdf: a print-ready PDF, text, shapes and lines drawn as vectors, pictures at the resolution asked for. |
| scale | number (0.25–3) | 1 | Pixels per page pixel when no print size is given; 2 for a sharper copy. |
| object | A print size instead of a pixel scale: unit mm, cm or in, the width (or the height — the other follows the design's proportions) and a resolution in dpi (300 when left out). A PNG or JPEG gets that many pixels to the inch and says so in its file; a PDF's page is exactly that size. E.g. { unit: "cm", width: 70.7, dpi: 300 } for a B1 poster. | ||
| quality | integer (1–100) | 1–100 for jpeg, webp and a PDF's pictures. | |
| transparent | boolean | png and webp: leave the page's own colour out. |
Multi-page documents: a carousel, a campaign, a batch gathered into one place.
List design projects
The user's design projects — a batch of posts kept as one document with pages, the way a Canva design has pages; the Studio opens a project and lays its pages out one under the other. Newest first, with how many pages each holds; with project one project in full, its pages in order (each a design id get_design, edit_design and render_design take). Free to call.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| project | id | One project, with its pages. | |
| search | string | Words to match against titles. | |
| limit | integer (1–100) | 30 |
Create a design project
A new multi-page project in the user's Design Studio: each page either a copy of a template or design with values poured into its slots (from + values), or one of the user's own designs moved in as a page (design). Pages keep the order given; a page is a page, so an infinite canvas cannot be one. Returns the project with its page ids — each a design that get_design, edit_design and render_design take — and its editor link; render_design_project paints every page. Free.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| title * | string (≤ 120 chars) | ||
| pages * | array of object (≤ 100 items) | ||
| render | boolean | false | Also queue a picture of every page. |
| format | "png" | "jpeg" | "webp" | "pdf" | "png" | |
| scale | number (1–3) | 1 |
Update a design project
Changes one of the user's projects: rename it; add pages at the end (a copy of a template with values, or one of the user's designs moved in — a page, never an infinite canvas); take pages out — detach_pages makes them standalone designs again, remove_pages deletes them; and set the order of every page with order (the page ids, first to last). Free.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| project * | id | ||
| title | string (≤ 120 chars) | ||
| add_pages | array of object (≤ 100 items) | ||
| detach_pages | array of id (≤ 100 items) | ||
| remove_pages | array of id (≤ 100 items) | ||
| order | array of id (≤ 100 items) | Every remaining page id, in order. |
Delete a design project
Removes one of the user's projects. Its pages go with it, unless keep_pages: true, which first makes every page a standalone design in the library. Cannot be undone.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| project * | id | ||
| keep_pages | boolean | false |
What the Bulk tab and the Create tab keep: batches of posts with their reviews, and conversations that continue turn by turn.
List bulk batches
The batches on the user's AI Post Editor Bulk tab — every run of many posts from one template, saved from its first keystroke: the template, the brief, the pasted rows, the model and the switches, and with batch one batch in full: every post's values, its picture prompts, whether it is selected, and what the review said of it. generate_design_variants with batch: true writes a run here so it appears on the tab. Free to call.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| batch | id | One batch, with its posts. | |
| limit | integer (1–100) | 30 |
List AI editor conversations
The briefs begun on the user's AI Post Editor Create tab, each kept as a conversation: the brief, the turns (what was asked, what the editor answered) and the design as it stands. Newest first; with conversation one in full — its turns, and its design's slots and layers. continue_design_conversation takes one further; generate_design with conversation: true starts one. Free to call.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| conversation | id | One conversation, in full. | |
| limit | integer (1–100) | 30 |
Continue an AI editor conversation
One more turn on a conversation from the user's AI Post Editor: the instruction is applied to the design as it stands ("make the headline shorter", "swap to a dark palette", "add a price tag bottom right"), the revised design becomes the conversation's current one, and both turns are kept — the tab shows them when the user opens it. sonnet costs 2 credits, opus 5. With save: true the result is also saved as a design of its own (render: true then queues its picture).
| Parameter | Type | Default | What it is |
|---|---|---|---|
| conversation * | id | ||
| instruction * | string (≤ 2000 chars) | ||
| model | "sonnet" | "opus" | The conversation's own model when omitted. | |
| save | boolean | false | Also save the revised design in the library. |
| title | string (≤ 120 chars) | ||
| render | boolean | false | |
| format | "png" | "jpeg" | "webp" | "pdf" | "png" | |
| scale | number (1–3) | 1 |
Delete an AI editor conversation
Removes a conversation from the user's AI Post Editor — its turns and the design it held. Designs saved to the library from it stay. Cannot be undone.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| conversation * | id |
ElevenLabs' Eleven v4, v4 Turbo and v3 reading scripts into the gallery, samples of every voice, favourites, voices of your own — cloned, designed or added by ID — a video re-voiced with the timing kept, and recordings turned into text.
List voice models and voices
The voice studio's catalogue: the text-to-speech models — Eleven v4 (the newest and the default), Eleven v4 Turbo (cheaper, near-instant) and Eleven v3, all ElevenLabs through fal — with their price per started 1,000 characters, and every preset voice with a line on how it sounds and the words it is found by (old, deep, warm, young, british, conversational…). search finds a voice by any of those. Signed in, it also lists the user's own voices (cloned from a recording, designed from a description, or added from ElevenLabs' Voice Library by ID — pass them to generate_speech as their custom:<id> key; a voice cloned or designed on MiniMax is read by MiniMax Speech 2.8 HD) and marks the voices they hearted with favorite: true; favorites_only narrows the list to those. Pass a model key and a voice id to generate_speech; voice_preview plays a sample; set_voice_favorite hearts one. Free.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| model | "elevenV4" | "elevenV4Turbo" | "elevenV3" | Only one model's voices. | |
| search | string | Words to find a voice by: name, age, accent, character, use. | |
| favorites_only | boolean | Only the voices the user hearted (needs sign-in). |
Hear a voice
A sample of one voice reading a fixed line — made once and kept, so it comes back at once for a voice anyone has heard before. Returns an MP3 link to play or pass on: public for a preset, signed for an hour for one of the user's own voices ("custom:<id>", whose sample is the line it read when it was made). Free: the script never changes and only the catalogue's voices are read.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| model * | "elevenV4" | "elevenV4Turbo" | "elevenV3" | ||
| voice * | string (≤ 120 chars) | A voice id from list_voices — a preset, or "custom:<id>". |
Generate speech (text to speech)
A script read aloud and saved to the user's gallery as an MP3 with a public link. Models (ElevenLabs, through fal): elevenV4 — Eleven v4, the newest and the default, the most natural read in 90+ languages; elevenV4Turbo — the same voices for less and nearly at once; elevenV3 — the previous model, for voices tuned on it. Voices: a preset from list_voices, or one of the user's own voices as "custom:<id>" (list_voices lists them when signed in) — a voice cloned or designed on MiniMax is read by MiniMax Speech 2.8 HD whatever model is named. Direction goes in square brackets in the text, in English whatever language the script is in: a delivery tag right before the words it colours ([warmly], [whispers], [excited], [sarcastic]) and a reaction where it happens ([laughs], [sighs], [gasps]) — about one per sentence or two; the brackets are never read out. Pronunciation goes in slashes as IPA. Optional stability (0 most expressive … 1 most even, default 0.5), similarity (v4 models only, default 0.75) and language (ISO 639-1; left out, it is read from the script). Brackets, stability and similarity are ElevenLabs' — a MiniMax voice takes only language, and its brackets are dropped. Up to 5,000 characters a call on v4 and Turbo, 6,000 on v3; longer scripts go in parts. Priced per started 1,000 characters at the model's rate (list_voices says it). Returns the file's links and the generation id — usable as reference audio for the video models that take a voice track, and filed into a Brain folder with save_output_to_brain.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| text * | string (≤ 6000 chars) | The script, in the language it should be read in. | |
| voice * | string (≤ 120 chars) | A preset voice id from list_voices, or one of the user's own as "custom:<id>". | |
| model | "elevenV4" | "elevenV4Turbo" | "elevenV3" | "elevenV4" | |
| stability | number (0–1) | 0 is the most expressive read, 1 the most even. Default 0.5. | |
| similarity | number (0–1) | How closely the read holds to the voice (v4 models only). Default 0.75. | |
| language | "hu" | "en" | "de" | "fr" | "es" | "it" | "pt" | "pl" | "ro" | "cs" | "sk" | "hr" | "nl" | "sv" | "tr" | "uk" | "ru" | "ja" | "ko" | "zh" | Pins the language (ISO 639-1); left out, it is read from the script. | |
| title | string (≤ 120 chars) | What the output is called in the gallery; the script's start otherwise. |
Heart a voice
Adds a voice to the user's favourites, or takes it off (favorite: false). Favourites lead the voice studio's lists and list_voices marks them (favorites_only narrows to them) — a way to keep the few voices a brand uses at hand. Takes a preset voice id or one of the user's own as "custom:<id>". Free.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| voice * | string (≤ 160 chars) | A preset voice id from list_voices, or "custom:<id>". | |
| favorite | boolean | true |
Add an ElevenLabs voice by ID
Adds any voice from ElevenLabs' Voice Library (elevenlabs.io → Voice Library → a voice → its voice ID, twenty letters and digits) to the user's own voices. It reads one sample line first (on Eleven v4, or v3 when v4 does not take it) — a voice fal cannot read is never added — and from then on it is "custom:<id>" in list_voices, generate_speech, voice_preview and change_video_voice. 1 credit, refunded if the voice does not work.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| voice_id * | string | The ElevenLabs voice ID. | |
| name * | string (≤ 60 chars) | What the voice is called in the studio. | |
| language | "hu" | "en" | "de" | "fr" | "es" | "it" | "pt" | "pl" | "ro" | "cs" | "sk" | "hr" | "nl" | "sv" | "tr" | "uk" | "ru" | "ja" | "ko" | "zh" | The language the voice is for (ISO 639-1). The sample line is read in it where one is written (hu, en, de, fr, es, it), in English otherwise. |
Clone a voice from a recording
Makes a voice of the user's own from one recording of a person speaking — their own voice, or one they have the speaker's permission for — and keeps it among their voices as "custom:<id>" for generate_speech, voice_preview and set_voice_favorite. ElevenLabs clones it (through fal), so it reads on Eleven v4, v4 Turbo and v3 and can re-voice a video with change_video_voice; should that be unavailable, MiniMax clones it instead and MiniMax Speech 2.8 HD reads it (not usable for change_video_voice). The recording: a sound file in the Brain (asset_id), a public https link (audio_url), or the bytes (audio_base64, mp3/wav/m4a), 10 seconds to 5 minutes of clear speech from one speaker — a minute is plenty — at most 20 MB. BEFORE calling, ask the user to confirm: "This is my own voice, or I have the speaker's explicit permission to clone it. I won't use it to impersonate anyone or to mislead." — and pass consent: true only when they have. 40 credits, refunded if the clone fails. Takes up to a minute.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| asset_id | id | A sound file in the Brain. | |
| audio_url | url | A public https link to the recording. | |
| audio_base64 | string | The recording's bytes, base64 (a data: URL works too). | |
| mime_type | string | What audio_base64 is, e.g. audio/mpeg. | |
| name * | string (≤ 60 chars) | What the voice is called in the studio. | |
| language | "hu" | "en" | "de" | "fr" | "es" | "it" | "pt" | "pl" | "ro" | "cs" | "sk" | "hr" | "nl" | "sv" | "tr" | "uk" | "ru" | "ja" | "ko" | "zh" | The language the recording is spoken in (ISO 639-1). | |
| consent * | true | The user confirmed it is their own voice or they have the speaker's explicit permission. |
Design a new voice from a description
Makes a brand-new voice — nobody's, made up from the words — from a description of how it should sound: age, gender, accent, character, pace, what it is for ("a deep, calm male narrator in his sixties, warm and unhurried, for documentaries"). MiniMax designs it and MiniMax Speech 2.8 HD reads it (Hungarian included); it is kept among the user's voices as "custom:<id>" for generate_speech, voice_preview and set_voice_favorite (not for change_video_voice, which takes ElevenLabs voices). It reads a first line — in its language where one is written (hu, en, de, fr, es, it), in English otherwise — returned as a sample link. 80 credits, refunded if it fails.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| description * | string (≤ 1000 chars) | How the voice should sound — the more specific, the closer. | |
| name * | string (≤ 60 chars) | What the voice is called in the studio. | |
| language | "hu" | "en" | "de" | "fr" | "es" | "it" | "pt" | "pl" | "ro" | "cs" | "sk" | "hr" | "nl" | "sv" | "tr" | "uk" | "ru" | "ja" | "ko" | "zh" | The language it will mostly speak (ISO 639-1). |
Change the voice in a video
Re-voices a video with ElevenLabs and puts the new voice back on the picture — the Higgsfield-style voice change. The speech is lifted out, ElevenLabs' voice changer reads it in the chosen voice (timing, breath and emotion kept, so it stays on the lips), and with keep_background (the default) the music and effects are mixed back under it. The source is one of: generation_id (a finished clip in the gallery — any generate_video output), asset_id (a video in the Brain) or video_url (a public https link). The voice is a preset from list_voices or one of the user's own ElevenLabs voices as "custom:<id>" (not one cloned or designed on MiniMax). Up to 5 minutes and 50 MB; every speaker gets the same new voice. 12 credits a minute (1 per started 5 seconds, at least 2), refunded if it fails. Returns at once with an id — the change runs in the background (a minute or two) and lands in the gallery as a new clip; get_voice_change follows it.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| generation_id | id | A finished video in the gallery. | |
| asset_id | id | A video in the Brain. | |
| video_url | url | A public https link to a video. | |
| voice * | string (≤ 120 chars) | A preset voice id from list_voices, or "custom:<id>". | |
| keep_background | boolean | true | Keep the music and effects under the new voice (default); false gives the voice alone. |
Follow a video voice change
Where a change_video_voice run stands — and each call moves it on: the step it is on, then the finished clip's gallery id and a signed link, or why it failed (credits are refunded then). A change takes a minute or two; there is no need to call this in a tight loop.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| id * | id | The id change_video_voice returned. |
Transcribe audio (speech to text)
A recording turned into text — a voice memo, an interview, a clip's soundtrack, a voiceover made with generate_speech — by the transcription model, and the transcript saved to the user's gallery. Point at a gallery output by generation_id (a tts output, or a clip), a public url, or send the bytes as audio_base64 (mp3, m4a, wav, webm, ogg, flac; up to 25 MB). 1 credit per started five minutes; when the recording's length is not given it is taken from its size at 128 kbit/s, so pass duration_seconds when it is known.
| Parameter | Type | Default | What it is |
|---|---|---|---|
| generation_id | id | A gallery output holding the recording. | |
| url | url | A public link to the recording. | |
| audio_base64 | string (≤ 34000000 chars) | The recording's bytes, base64 (or a data URL). | |
| mime_type | string | What audio_base64 is, e.g. audio/mpeg. | |
| duration_seconds | number (0–3600) | The recording's length, for the price. |
MCP access is included on every plan. Connect once; every tool runs as you.
Start creating free
Social scheduling
Posts to Instagram, TikTok, YouTube, Facebook, X, LinkedIn, Threads, Pinterest and Bluesky through Post for Me: connecting accounts, scheduling posts with pictures and clips from the Brain or the gallery, following how they went, calling them off, and add-on posts past the ones the plan includes each billing month.
list_social_accounts
read-onlyList connected social accounts
The social accounts the user has connected for scheduling posts (Instagram, TikTok, YouTube, Facebook, X, LinkedIn, Threads, Pinterest, Bluesky) — each with the id schedule_social_post takes, its platform, username and status — and the posts by billing period — a month from the day the user's plan renews, as its credits do: this period and every later one with posts in it: published, scheduled and failed (all counted), what did not count, and what is left of each period's allowance — with how many posts are still to go out and the accounts they go to. Posting comes with a paid plan: on Free (planNeeded) no account connects and no post goes out — point the user to /pricing. A user can have several accounts on one platform (three Instagram accounts, say); each has its own id, and one post can go to any of them. An account listed as disconnected (its sign-in lapsed, or it was disconnected) takes no posts until it is connected again with connect_social_account. A post counts once for each account it goes to. Call it first, before schedule_social_post; an account missing from it is connected with connect_social_account. Free to call.
connect_social_account
Connect a social account
The link the user opens to connect one of their social accounts — Instagram, TikTok, YouTube, Facebook, X, LinkedIn, Threads, Pinterest, Bluesky — so posts can be scheduled to it. They sign in on the platform's own page (through Post for Me, the service that posts for Lumen Brain), approve, and land back on Lumen Brain. Show them the link to open, and once they say they are back, call list_social_accounts: the new account is listed with the id schedule_social_post takes. SEVERAL ACCOUNTS ON ONE PLATFORM (three Instagram accounts, say) are fine, each connected once, with no need to sign out: the platform's page offers the account signed in on that browser, so for another one the user switches account on that page ("Switch accounts" / "Not you?"), or opens the link in a private window and signs in there. For Instagram, instagram_via: "facebook" signs in through Facebook instead, where every business or creator account linked to the user's Facebook Pages can be ticked at once — the quickest way to connect several. Instagram posts only to business or creator accounts. A disconnected account is connected again the same way, signing in as that account. Bluesky has no sign-in page: it connects with the account's handle and an app password (made in Bluesky under Settings → Privacy and security → App passwords — never the account's own password). The Social page at https://lumenbrain.app/social is the safer place to enter one, since it does not pass through the chat; pass bluesky_handle and bluesky_app_password here only when the user chooses to give them. Free to call.
schedule_social_post
Schedule a social post
Schedules a post to one or more of the user's connected social accounts, or sends it now, through Post for Me: a caption, pictures or a clip from the Brain, the gallery or a link, and the time it goes out. A post goes out in public under the user's name and cannot be taken back once it is out, so confirm the caption, the accounts and the time with the user first. Call list_social_accounts first for the account ids (connect_social_account connects a missing one). Give scheduled_at with the UTC offset in force in the user's time zone on the day it goes out (daylight saving changes it) — e.g. 2026-11-03T09:00:00+01:00 — and ask the user for their time zone when it is not known: a time without an offset is refused. Left out, the post goes out now; it can be scheduled up to 90 days ahead. Each platform's rules are checked before anything is counted: captions up to X 280, Bluesky 300, Threads 500, Pinterest 500, Instagram 2,200, TikTok 2,200, LinkedIn 3,000, YouTube 5,000 and Facebook 63,206 characters (the shortest among the chosen accounts' platforms applies); Instagram, TikTok and Pinterest need a picture or a clip, YouTube a clip; a clip goes out on its own; Pinterest needs a board. TIKTOK PHOTOS: TikTok lays a photo post's pictures across the phone's full width with their middle a little above the screen's, so on a 1080 × 1920 picture about the top 230 px sit under the phone's status bar and the next 110 or so under TikTok's own bar — keep words and logos at least 360 px from the top, above about 1,590 px (the dots and the caption sit below that) and out of the right-hand 150 px from about 960 px down, where TikTok's buttons are (measured on an iPhone; other phones shift it a little). Returns the post id, when it goes out (in UTC and as given), the accounts, and how many posts it used and how many its billing period has left; list_social_posts shows how it went on each account, reschedule_social_post moves it to another time and cancel_social_post calls it off before it goes out. COUNTING: a post to N accounts counts as N posts, in the billing period it goes out in — a month from the day the user's plan renews, as its credits do (list_social_accounts names the period). Each paid plan includes a number of posts a period at no credit cost (list_social_accounts shows how many are left); Free includes none, so on Free this tool refuses and the user needs a paid plan (/pricing). Past the period's posts this tool refuses, and the user can add 100 more for 30 credits with add_social_post_pack — ask them before buying one — or post to fewer accounts. A post Post for Me refuses gives its posts back.
list_social_posts
read-onlyList social posts
The user's social posts, newest first: each with its id, its status (scheduled, publishing, posted, partial, failed, rejected or canceled), when it goes or went out, a caption excerpt, the accounts it goes to and — once it is out — the link to it or the error on each account. filter picks the upcoming ones (still to go out; the default), the finished ones, or all. Posts whose time has come are brought up to date with Post for Me as they are read, so a post sent a minute ago shows how it went when this is called again. A scheduled post's id goes to reschedule_social_post, split_social_post or cancel_social_post; a failed or partly posted one's to retry_social_post. A post YouTube turned down for its daily upload limit gives that post back and goes again on its own just after the limit starts over — the listing says when. Free to call.
reschedule_social_post
Move a scheduled social post to another time
Moves a scheduled social post to another time before it goes out — everything else about it stays as it is. Give scheduled_at with the UTC offset in force in the user's time zone on the day it goes out (daylight saving changes it), e.g. 2026-11-03T09:00:00+01:00 — ask the user for their time zone when it is not known: a time without an offset is refused. The new time must be a few minutes ahead or more, and up to 90 days ahead. A post moved into another billing period (a month from the day the user's plan renews) counts there instead, and is refused when that period has too few posts left (add_social_post_pack adds more, once the user agrees to spend the credits); the period it leaves gets its posts back. Only a post still waiting for its time can be moved — not one going out within the minute, nor one already out or called off. Takes the post id from schedule_social_post or list_social_posts. Free to call.
split_social_post
Send one platform of a scheduled post at another time
Takes one platform's accounts out of a scheduled post into a post of their own at another time, the rest still going at the post's own time — for YouTube above all: its daily upload limit is Post for Me's, shared by everyone posting through it, starts over at midnight Pacific time and is often used up by the afternoon there, so a YouTube upload early in its day goes through where an evening one in Europe is turned down. Give scheduled_at with the UTC offset in force in the user's time zone that day, e.g. 2026-11-03T09:15:00+01:00. The new post counts that platform's accounts in the billing period it goes out in and the post they leave gives them back, so within a period the count stays the same. A post going to that platform alone simply moves, as reschedule_social_post moves it. Only a post still waiting for its time can be split. Takes the post id from list_social_posts. Free to call.
cancel_social_post
destructiveCall off a scheduled social post
Calls off a scheduled social post before it goes out: it is taken off Post for Me and its posts go back to the billing period they were counted in. Only a post still waiting for its time can be called off — not one going out within the minute, nor one already out. Takes the post id from schedule_social_post or list_social_posts. Free to call.
share_social_post
Share a social post's preview by link
Makes a link to a social post's preview — the post as each platform it goes to will show it, and when it goes out — that anyone can open without an account: for a client to look over a post made for them, before or after it goes out. Nothing else of the account goes with it. share=true makes the link (or returns the one the post already has); share=false turns it off, and the old link stops working. Takes the post id from schedule_social_post or list_social_posts. Free to call.
share_social_calendar
Share the whole posting calendar by link
Makes one link to the user's whole posting calendar that anyone can open without an account: a month at a time, every post scheduled and every one that went out, each opening as its platforms will show it — for a client or a team to follow the schedule. Posts called off or that did not go out are left off, and nothing else of the account goes with it. Posts scheduled later show on it by themselves. share=true makes the link (or returns the one the calendar already has); share=false turns it off, and the old link stops working — sharing again makes a new one. For a single post's preview, use share_social_post. Free to call.
retry_social_post
Send a failed social post again
Sends a post that failed on every account, or went out on only some, again — to the accounts it missed, as a new post with the same caption, pictures or clip and settings: now, or at scheduled_at. It goes out in public under the user's name, so confirm with the user first. Give scheduled_at with the UTC offset in force in the user's time zone on the day it goes out, e.g. 2026-11-03T09:00:00+01:00, or leave it out to send now. A platform that turned the post down for its posting limit (list_social_posts says so) takes it better later: YouTube's daily upload limit starts over at midnight Pacific time — and where list_social_posts says a post goes to YouTube again on its own, this sends it only to the other accounts it missed. The new post counts as one post per account in the billing period it goes out in (a month from the day the user's plan renews), like any post; past the period's posts it is refused, and add_social_post_pack adds more once the user agrees to spend the credits. A post Post for Me refused (rejected) is not sent again as it is — change it and use schedule_social_post. Takes the post id from list_social_posts. Free to call.
add_social_post_pack
Add social posts
Buys add-on social posts for a billing period with the user's credits: 100 posts for 30 credits a pack, up to 20 packs at a time, for this period or one of the next three. A billing period is a month from the day the user's plan renews (list_social_accounts names it, e.g. Nov 17 – Dec 16); posts count in the period they go out in, and a pack's posts are for its period only — they do not carry over. Free buys none: posting comes with a paid plan. This spends credits: only call it once the user has agreed to spend them — when schedule_social_post refuses because the period's posts are used up, tell the user what a pack costs and ask first. Returns the period's posts after: what the plan includes, what was bought, what is used and what is left.