An n8n community node for Video API — AI-powered video and audio processing.
A single Video API node exposes the whole service, organized by Resource (Video, Audio, Render, File, Job) and Operation. Every operation is also searchable by name in the n8n nodes panel.
In n8n (recommended): Settings → Community Nodes → Install → @mobiuscode/n8n-nodes-videoapi. See the community node install guide.
Manual / self-hosted: install the package into your n8n custom folder:
cd ~/.n8n/nodes # create it if it doesn't exist
npm install @mobiuscode/n8n-nodes-videoapiGet an API key from video-api.io, then create Video API credentials in n8n:
- API Key — sent as the
x-api-keyheader. - Base URL — defaults to
https://api.video-api.io; change it only for a self-hosted/local instance.
n8n validates the key when you save the credential.
Most media operations create an asynchronous job. Every job-creating operation shares a Job Options group:
| Option | Default | Description |
|---|---|---|
| Wait for Completion | true |
Poll the job until it finishes and return the completed job JSON (its result field is the output file ID). Turn off to return the raw JobResponse (with the job id) and poll yourself with Job → Get. |
| Download Result as Binary | false |
Also download the result file and attach it as binary (data). Off by default — fetch the file anytime with File → Download instead. Only applies while Wait for Completion is on. |
| Create as Draft | false |
Create the job without queueing it. Run it later with Job → Queue. (Draft implies "don't wait".) |
| Polling Interval (Ms) | 2000 |
How often to check job status while waiting. |
| Timeout (Ms) | 1800000 |
Maximum time to wait for completion (default 30 min). |
A job's output is a file ID (result). To chain operations, pass that ID into the next operation's file input — the file stays on Video API's servers (upload once, download once at the end).
The three AI operations are self-contained: their Input Mode accepts File IDs, Binary (auto-uploaded), or a URL (auto-pulled), so a single node can take a source and return a finished video. The URL/Pull download must point to a direct media file — a YouTube or web-page URL won't work, and only download content you have the rights to.
| Operation | Description |
|---|---|
| AI Cut | Create a new video by AI-cutting sequences from one or more videos, guided by a prompt. |
| AI Sequence Extraction | Extract video sequences from a video by prompt. |
| AI Social Media Compose | Compose a social-media video with AI voice-over, music and subtitles. |
| Caption | Burn given subtitle text into a video. |
| Cut | Cut and merge specific video sequences. |
| Extract Audio | Extract the audio track from a video. |
| Generate Preview | Generate preview images from a video. |
| Get Info | Get video metadata (resolution, duration, FPS…). |
| Mix In Audio | Mix an audio file into a video (kept alongside existing audio). |
| Overlay | Overlay videos/images over a background video. |
| Replace Audio | Replace a video's audio with a given audio file. |
| Subtitles | Generate and render subtitles into a video. |
| Transcode | Transcode a video to a different format/resolution. |
Non-AI operations take file IDs directly — get one with File → Upload or File → Pull first.
| Operation | Description |
|---|---|
| Generate From Script | Text-to-speech voice-over from a script (30 voices). |
| Get Info | Get audio metadata (duration). |
| Transcode | Convert audio to FLAC / AAC / AAC HQ / MP3. |
| Operation | Description |
|---|---|
| Subtitles | Render subtitles for an audio or video file (WebM or ZIP). |
| Operation | Description |
|---|---|
| Upload | Upload binary data; returns the file id. |
| Pull | Download a direct media URL into Video API; returns the resulting file id. |
| Download | Download a stored file by ID as binary. |
| Download Zip | Download multiple files (comma-separated IDs) as a ZIP. |
| Get Metadata | Get metadata for a stored file. |
| Get Download Token | Get a short-lived download token. |
| Delete | Permanently delete a stored file. |
| Operation | Description |
|---|---|
| Get | Get all information about a job. |
| List | List jobs (with page/size/status filters). |
| Queue | Transition a DRAFT job to SCHEDULED. |
| Delete Draft | Delete a DRAFT job. |
The examples/ folder has ready-to-import workflows for the AI operations. Each is a single Video API node (Input Mode = URL, Wait for Completion = on, Download Result as Binary = on) that returns the finished file as binary (data) — import one, select your Video API credential, and run:
ai-cut.json— AI-cut a video into a short clip from a prompt.ai-sequence-extraction.json— extract sequences from a video by prompt.ai-social-media-compose.json— compose a portrait social-media video with voice-over, music and subtitles.
For long videos, swap Wait-on for the poll-loop in Tips & gotchas below.
The ID fields default to ={{ $json.result }}, which grabs the file ID from the previous node automatically. Where the ID lives depends on the upstream node:
- After a job (AI Cut, Transcode, Pull…) or Job → Get → it's in
result. Itsidis the job ID, not the file. - After File → Upload or Get Metadata → it's in
id(noresult). Use={{ $json.id }}there. - After File → Download → it's in
fileId.
Override the field anytime; the default is just a convenience for the common case.
These need two different files (e.g. a video and an audio). $json only refers to the one node immediately before, so the auto-default can only fill one field correctly — if you leave both on the default they'd point at the same file.
Set the second input from its own source using a named-node reference:
Replace Audio
Video File ID = {{ $('Pull Video').item.json.result }}
Audio File ID = {{ $('Generate From Script').item.json.result }}
(Use .id instead of .result when the referenced node is a File → Upload.)
Get Info outputs only metadata (no id/result), so it breaks an ID chain. Put it on a side branch off the node whose file you want to inspect, not in the middle of the chain.
Wait for Completion = on blocks the node until the job finishes (capped by the Timeout, default 30 min) and ties up an n8n worker. For big AI renders, use fire-and-forget + a poll loop instead:
File: Pull/Upload → AI Cut (Input Mode = File IDs, Wait OFF) → Job: Get ◄─────┐
│ │
▼ │
IF status==DONE? │
┌──┴──┐ │
true│ │false │
▼ ▼ │
File: Download Wait 30s ────┘
- AI Cut → Wait OFF returns the job (
id) immediately, no blocking, no binary. - Job → Get (
Job ID = {{ $json.id }}) → IF ({{ $json.status }} == DONE) → on false a Wait node (15–30 s) loops back to Job → Get. - On true, File → Download (
{{ $json.result }}) fetches the finished video. - Pre-resolve the source to a file ID (Pull/Upload) and feed AI Cut via Input Mode = File IDs — then nothing inside the AI node blocks (URL/Binary modes still pull/upload the source synchronously first).
- Robustness: also branch on
{{ $json.errorReason }} != NONEso a failed job doesn't loop forever, and optionally stop after N polls.
- Binary output is opt-in — jobs return JSON by default (the file ID is in
result). Turn on Download Result as Binary to also get the file asdata, or fetch it anytime with File → Download. - Create as Draft = on doesn't queue the job, so the node won't wait — run it later with Job → Queue.
Operations consume credits from your Video API account (e.g. AI Cut: 1 credit per 60 s of input; voice generation: 1 credit per 1000 tokens). The returned JobResponse includes estimatedCosts and paidCredits.
git clone https://github.com/mobiuscode-com/n8n-nodes-videoapi.git
cd n8n-nodes-videoapi
npm install
npm run build # tsc + copies icons into dist/To test against a local n8n, point n8n at the build output and start it:
export N8N_CUSTOM_EXTENSIONS="$(pwd)/dist"
n8n start # http://localhost:5678Rebuild and restart n8n after changes. Source lives in nodes/ and credentials/; dist/ is generated.
Publishing is automated with semantic-release: every push to main derives the next version from the commit messages (Conventional Commits), tags it, creates a GitHub release and publishes to npm.
fix: …→ patch release (1.0.0 → 1.0.1)feat: …→ minor release (1.0.0 → 1.1.0)feat!: …orBREAKING CHANGE:footer → major release (→ 2.0.0)chore: …,docs: …, … → no release