Reference

All 48 workflow nodes.

Aifluence builds videos out of 48 node types on a canvas. A node is one step: fetch something, ask the AI, generate an image, speak a line, render, check, publish. Every node states what it takes and what it hands to the next step, so a workflow can be read top to bottom instead of guessed at.

This page lists all of them, grouped the way they appear in the app. The third column is the part people usually have to find out by trial and error: the fields a node adds to the record it passes on, which is what the next step can reference.

Triggers · 4

What starts a workflow — by hand, on a schedule, from a feed, or from another program.

Node	What it does	Hands on
Start `trigger.manual`	Also passes on runFolder — one folder for everything this run makes — and recentTitles, what this workflow made before.	`startedAt`, `recentTitles`, `runFolder`
Schedule `trigger.interval`	Runs only while “Automatic” is on for the workflow — but then regardless of whether this panel is open. Settings: mode, seconds, time, days	`firedAt`, `recentTitles`, `runFolder`
Started from outside `trigger.webhook`	Lets another program start this workflow: a Stream Deck button, a script that runs after your edit, a service like Zapier. Settings: path, method, secret	`body`, `query`, `method`, `path`, `receivedAt`, `recentTitles`, `runFolder`
New in a feed `feed.rss`	Any blog, news site or podcast that publishes a feed — the address usually ends in /rss, /feed or . Settings: url, onlyNew, limit	`title`, `url`, `published`, `summary`, `id`

Flow · 3

Branching, waiting and looping.

Node	What it does	Hands on
Condition `flow.if`	Settings: expression	der Datensatz geht unveraendert weiter
Wait `flow.wait`	Settings: seconds	der Datensatz geht unveraendert weiter
Split `flow.loop`	Settings: path	der Datensatz geht unveraendert weiter

Data · 11

Fetching, reshaping, filtering and remembering records as they travel between steps.

Node	What it does	Hands on
HTTP request `data.http`	Settings: method, url, headers, body	`status`, `body`, `json`
Connected API `data.api`	The access token stays in the main process and is only attached after leaving the panel — no secret is ever written into the workflow. Settings: service, method, path, body	`status`, `body`, `json`
Run a script `script.run`	The assistant writes the script into the workflow’s project folder. Settings: interpreter, file, args, secrets	whatever the script prints as a JSON line
Code `data.code`	Available: items, $json, $item, $index, $now, $seed. Settings: code	was der Code zurueckgibt
Set fields `data.set`	Settings: fields, keepOld	the fields from the configured JSON
Split a list `data.split`	One record holding a list becomes many. Settings: field, keepOld	`index`, `total`, the fields of the list entry
Collect into one `data.collect`	The counterpart to "Split a list": many records become ONE, holding them all as a list. Settings: field, keep, carry	`total`, the field named by "field" (default: scenes), holding every incoming record
Merge branches `data.merge`	Use this where parallel paths meet — for example, one branch generates images while another creates audio. Settings: mode, key, includeUnmatched, conflicts	die zusammengeführten Felder aller parallelen Zweige
Keep only some `data.filter`	Like the condition, but without a second branch: what does not match is dropped here. Settings: expression	the record passes through unchanged, just fewer of them
Sort and keep the first `data.pick`	For "take the three best clips": sort by a value the step before produced, then keep the front of the list. Settings: value, order, count	the record passes through unchanged, sorted and trimmed
Skip what was already handled `data.dedupe`	Lets an entry through the first time and never again. Settings: identity, scope, limit	the record passes through unchanged, minus anything seen before

Media · 22

The production itself: analysis, script, images, voice, rendering, checking and publishing.

Node	What it does	Hands on
File `input.file`	Settings: path, field, askEveryRun	`file (bzw. der eingestellte Feldname)`
Several files `input.files`	Produces one record per file, sorted alphabetically — plus index and total. Settings: source, folder, only, field, askEveryRun	`file, index, total — ein Datensatz je Datei`
Cut video `media.ffmpeg`	Settings: preset, input, output, start, duration, args	`output`, `file`
Transcribe `ai.transcribe`	Local needs Python and faster-whisper — the assistant can set both up. Settings: mode, file, language, device	`transcript`, `segments`
Analyse a video visually `ai.video`	Checks fresh video frames with timestamps on every run. Settings: file, prompt, maxFrames, outputKey, json	the field named by outputKey (default: videoAnalysis), `videoMetadata`
AI analysis `ai.claude`	Uses your Aifluence account automatically. Settings: system, prompt, outputKey, json, schema	the field named by outputKey (default: analyse)
Generate image `ai.image`	Generated securely through your Aifluence account. Settings: prompt, initImage, width, height, steps, outPath	`image`
Speech `audio.speak`	Natural speech through your Aifluence account. Settings: text, voice, rate, wordsPerCaption, outPath	`audio`, `seconds`, `segments`
Sound effect at moments `video.sfx`	Mixes one file in at every second listed, leaving the existing sound alone. Settings: video, effect, times, volume, outPath	`video`, `file`
Add music `video.music`	Loops the song under existing audio, or adds it to a silent video. Settings: video, music, volume, outPath	`video`
Story over background video `video.background`	Loops and crops one reusable video locally, then replaces its sound with the narration. Settings: background, audio, seconds, start, size, focus, outPath	`video`, `file`, `backgroundVideo`, `seconds`
Burn in captions `video.captions`	Settings: video, segments, clipStart, clipDuration, maxWords, position, font, fontSize, bold, outline, shadow, marginV, outPath	`video`, `file`, `captionCount`
Check the video `video.validate`	Stops the workflow with a clear message if file, length, aspect ratio or audio are wrong. Settings: file, aspect, minSeconds, maxSeconds, requireAudio	a check result; the record is otherwise unchanged
Video from scenes `video.timeline`	Gathers every incoming record into ONE video — or reads a ready-made list if a single record carries scenes: [{image, audio, caption, seconds, camera}]. Settings: image, audio, caption, seconds, camera, fade, size, fps, captionPosition, captionSize, captionOutline, captionLines, captionWidth, outPath	`video`, `scenes`, `seconds`, `sceneStarts`
Still from a video `video.frame`	Without a time it grabs at about 35% of the runtime — the opening is usually a logo or a black frame. Settings: video, at, outPath	`image`, `frame`, `frameAt`
Either-or card `image.quiz`	One card per incoming record. Settings: topImage, bottomImage, topText, bottomText, topPercent, bottomPercent, badge, topColor, bottomColor, size, outPath	`image`, `card`
Build a thumbnail `image.thumbnail`	The background can be a still from the video or a generated image. Settings: image, text, subtext, position, accent, size, outPath	`thumbnail`
Build a video `video.build`	Image duration follows the length of the audio track, so the two match. Settings: images, audio, size, fps, outPath	`video`
Prepare for platforms `publish.variants`	One record per platform, with the right crop and platform-appropriate text. Settings: file, networks, title, caption, hashtags, convert	`network`, `video`, `file`, `title`, `caption`, `hashtags`, `format`
New YouTube videos `youtube.feed`	Uses the public RSS feed — no API key needed. Settings: channelId, onlyNew	`videoId`, `title`, `url`, `published`
Publish on YouTube `youtube.upload`	Uploads the video directly through the connected YouTube account. Settings: file, idempotencyKey, allowDuplicates, title, description, tags, thumbnail, privacyStatus, publishAt, categoryId, madeForKids, containsSyntheticMedia, notifySubscribers, language	`uploaded`, `alreadyPublished`, `idempotencyProtected`, `network`, `videoId`, `url`, `studioUrl`, `title`, `privacyStatus`, `publishAt`, `thumbnailUploaded`
Prepare an upload `upload.prepare`	Opens the upload page in a panel, copies the file path to the clipboard and puts the text ready. Settings: network, file, caption	`uploadReady`, `network`, `file`, `caption`

Workspace · 4

Driving the browser panels on the canvas.

Node	What it does	Hands on
Open a panel `panel.open`	Settings: network, url	der Datensatz geht unveraendert weiter
Panel opens URL `panel.navigate`	Settings: panelId, url, wait	der Datensatz geht unveraendert weiter
Read the page `panel.extract`	Settings: panelId, code	whatever the code on the page returns
Tidy up `panel.tidy`		der Datensatz geht unveraendert weiter

System · 4

The computer around the workflow — programs, OBS, files and notifications.

Node	What it does	Hands on
Start a program `system.app`	Settings: exe, args	der Datensatz geht unveraendert weiter
OBS action `system.obs`	Settings: action, scene, url, password	`obs`
Notification `system.notify`	Settings: title, body	der Datensatz geht unveraendert weiter
Write a file `system.file`	Settings: path, content, append	`path`

See them working

Aifluence is a Windows 10/11 desktop app. The assistant wires these nodes for you from a prompt or a reference video — you keep every one of them editable.

Download for Windows