OpenMontage

Open-source, agentic video production studio that plugs into your AI coding assistant.

Free· Free and open-source (AGPL-3.0). Runs at $0 with free providers (Piper TTS, Remotion, Archive.org/NASA/Wikimedia footage). Optional paid provider costs typically run $0.15-$3 per video when using Kling, Runway Gen-4, Veo 3, ElevenLabs, Suno, FLUX, etc. Configurable budget cap defaults to $10 per run.VideoProvider-agnostic orchestrator: FLUX, Imagen, GPT Image 2, Stable Diffusion, Kling, Runway Gen-4, Veo 3, HeyGen, ElevenLabs, Suno; driven by any AI coding assistant.

Visit website →

Best for

Technical creators, indie studios, and dev-savvy marketing teams who already live inside Claude Code / Cursor / Copilot and want a scriptable, provider-agnostic video pipeline they can extend and self-host.

Skip if

Non-technical users who want a click-and-render web app, teams that need a closed-source-friendly licence, or anyone expecting a single "generate video" button without configuring providers or reviewing intermediate stages.

OpenMontage is an open-source, agentic video production system that turns an AI coding assistant (Claude Code, Cursor, Copilot, Windsurf, or Codex) into a full end-to-end video studio. Instead of generating a single clip from a prompt, it orchestrates the entire pipeline: research, scripting, scene planning, asset generation, editing, composition, and post-render QA. The distinguishing move is that it produces actually-edited videos from real motion footage - sourced from Kling, Runway Gen-4, Veo 3, HeyGen, MiniMax, and local GPU models, or from free archives like Pexels, Pixabay, Archive.org and Wikimedia - rather than animating still images.

The project ships 12 specialised pipelines (animated explainer, talking head, documentary montage, cinematic, avatar spokesperson, clip factory, screen demo, podcast repurpose, localisation and dub, hybrid, animation, trailer-style) plus 100+ tools spanning video generation (15 providers), image generation (11 providers including FLUX, Imagen, GPT Image 2, Recraft, Stable Diffusion), text-to-speech (ElevenLabs, Google TTS, OpenAI TTS, Kling, and offline Piper), music (Suno, ElevenLabs Music/SFX), and FFmpeg-based post. Composition is handled by Remotion (React), HyperFrames (HTML/CSS/GSAP), and raw FFmpeg. Around 700+ agent skill and knowledge files guide the coding assistant through each stage, with approval gates at proposal, script, assets, and publish steps, plus post-render self-review (ffprobe checks, frame sampling, slideshow-risk scoring, budget governance).

Typical workflow: you describe a video ("60-second animated explainer about neural networks"), the agent proposes an approach and budget, drafts a script with live web research, plans scenes, generates or fetches assets, assembles the edit, and runs automated QA before final render. Platform profiles for YouTube (including Shorts and 4K), Instagram Reels/Feed, TikTok, LinkedIn, and cinematic formats are built in. API keys are optional - functionality gracefully scales up with whichever providers you configure.

Editor's take

This is the most ambitious open-source video-agent project I've seen: it treats video production as an orchestration problem, not a generation problem, and the approval gates plus ffprobe-based self-review are a genuine step up from the usual "prompt in, mystery clip out" tools. The catch is that you really do need to be comfortable in a coding editor and willing to shop for provider keys - but if you are, the ceiling is far higher than any hosted competitor.

— The AI Tool Bible editorial team

Pros

✅ Genuine end-to-end pipeline (research to final render) rather than single-shot clip generation
✅ Provider-agnostic - 15 video, 11 image, 5 TTS providers plus free stock sources; scales with what you have keys for
✅ Runs at $0 using Piper TTS, Remotion, and free archive footage; paid providers are opt-in per run
✅ Multi-point human approval gates and automated post-render QA (ffprobe, slideshow-risk scoring, budget caps)
✅ 12 opinionated pipelines cover most common formats (explainer, talking head, documentary, cinematic, shorts)
✅ AGPL-3.0 open-source with extensible tool/pipeline modules and ~45k GitHub stars of community traction
✅ Built-in platform profiles for YouTube, Reels, TikTok, LinkedIn with correct aspect ratios and specs
✅ Works inside your existing coding editor - no separate SaaS dashboard to learn

Cons

⚠️ Requires an AI coding assistant subscription (Claude Code, Cursor, Copilot, etc.) - it is not a standalone app
⚠️ Local install with Python 3.10+, Node 18+, and FFmpeg prerequisites; not point-and-click for non-technical users
⚠️ Higher-quality outputs still depend on paid third-party APIs (Kling, Runway, ElevenLabs, Suno) that bill separately
⚠️ AGPL-3.0 copyleft makes it awkward to embed inside closed-source commercial products
⚠️ Agentic orchestration means longer runtimes and more moving parts than a single-prompt video generator
⚠️ Documentation and pipelines are evolving quickly; edge cases may require reading the source or joining Discussions

Use cases

AI-generated YouTube explainer videosTalking-head shorts with cloned or synthetic voicesDocumentary-style montages from archival footageProduct demo and screen-recording editsPodcast-to-video repurposing with captionsLocalised and dubbed versions of existing videosAnimated cinematic trailers and title sequencesAvatar spokesperson videos via HeyGen or similarAutomated Instagram Reels and TikTok clip factoriesProgrammatic video composition with Remotion