OPEN SOURCE · VIDEO AUTOMATION · KNOWLEDGE

Meet Vidyom.

Preserve originals. Find meaningful moments. Create new videos. Publish everywhere — with a programmable, extensible workflow engine.

Explore the documentation →View architecture

Architecture and implementation specification · Not yet a production release

From original recordings to meaningful clips

Overview of the two Vidyom workflows: original ingestion builds a searchable library, and clip creation finds key moments, renders new videos and publishes them across channels.

Illustrative overview of the two proposed workflows. See the specified stages →

Designed for an open media ecosystem

◈

Workflow-native

Design repeatable video pipelines in JSON, inspired by n8n. No code required to compose existing nodes.

⌘

CLI + daemon

Control persistent background jobs locally or securely across your private network.

✧

AI-assisted clipping

Find complete stories from natural-language prompts and render independent derivative assets.

▦

Portable media archive

Keep canonical assets, metadata, captions, and lineage independent of platforms.

⇄

Connector ecosystem

Use pluggable storage, publishing, AI, catalog, and observability providers.

◇

Secure by design

Keyring-backed secrets, private networking, policy-aware plugins, and auditable jobs.

One library. Unlimited possibilities.

$ vidyom original import yoga-class.mp4 --workflow ingest-original $ vidyom clip create --source ORI-001 --prompt "Explain Ujjayi breathing" $ vidyom job logs JOB-001 --follow

Illustrative proposed commands. The CLI and daemon are specified but not implemented yet.