OpenMontage — Turn Your AI Coding Assistant Into a Full Video Production Studio
What Is OpenMontage?
OpenMontage is the first open-source, agentic video production system. Instead of using a dedicated GUI or SaaS platform, you describe what you want in plain language, and your AI coding assistant (Claude Code, Cursor, Copilot, Windsurf, Codex) orchestrates the entire production pipeline — research, scripting, asset generation, editing, and rendering.
Created by Calesthio (Seattle-based, Microsoft engineer), OpenMontage has skyrocketed to 22,600+ GitHub stars since its debut in March 2026. It's currently #1 on GitHub Trending and represents a paradigm shift: your code editor becomes your video studio.
Why Is It Trending?
OpenMontage solves a fundamental problem: AI video tools mostly give you a single clip from a prompt. OpenMontage gives you an end-to-end production pipeline — the same structured process a real production team follows, automated by your AI agent.
Key factors driving its popularity:
- 12 production pipelines — explainers, talking heads, screen demos, cinematic trailers, animations, podcast repurposing, localization/dub, documentary montages, and more
- 52+ production tools — spanning video generation, image creation, TTS, music, audio mixing, subtitles, enhancement, and analysis
- 400+ agent skills — production skills, pipeline directors, creative techniques, and quality checklists
- Reference-driven creation — paste a video you like and the agent turns it into a differentiated production plan
- Real-footage documentary creation — builds actual edited videos from free/open motion footage, not just Ken Burns over images
- No vendor lock-in — swap providers freely via a scored selector that ranks every tool across 7 dimensions
- Budget governance — estimate before execution, spend caps, per-action approval thresholds. No surprise bills.
Prerequisites
- Python 3.10+
- FFmpeg —
sudo apt install ffmpegorbrew install ffmpeg - Node.js 18+
- An AI coding assistant — Claude Code, Cursor, Copilot, Windsurf, or Codex
- Optional API keys for cloud providers (see below)
Quick Start
Installation
git clone https://github.com/calesthio/OpenMontage.git
cd OpenMontage
make setupIf make is not available:
pip install -r requirements.txt
cd remotion-composer && npm install && cd ..
pip install piper-tts
cp .env.example .envOn Windows, if npm install fails with ERR_INVALID_ARG_TYPE, use npx --yes npm install instead.
Your First Video
Open the project in your AI coding assistant and tell it:
"Make a 45-second animated explainer about why the sky is blue"That's it. The agent researches your topic with live web search, generates AI images, writes and narrates the script, finds royalty-free background music automatically, burns in word-level subtitles, and renders the final video.
Before you see anything, the system runs a multi-point self-review — ffprobe validation, frame sampling, audio level analysis, delivery promise verification, and subtitle checks.
Adding API Keys (Optional)
Edit .env — every key is optional, add what you have:
# Image + video gateway:
FAL_KEY=your-key # FLUX images + Google Veo, Kling, MiniMax video
# Free stock media:
PEXELS_API_KEY=*** # Free stock footage and images
UNSPLASH_ACCESS_KEY=your-key # Free stock images
# Voice & images:
ELEVENLABS_API_KEY=*** # Premium TTS, AI music, sound effects
OPENAI_API_KEY=*** # OpenAI TTS, DALL-E 3 images
# More video providers:
HEYGEN_API_KEY=*** # HeyGen gateway
RUNWAY_API_KEY=*** # Runway Gen-4 directHave a GPU? Unlock free local video generation:
make install-gpu
# Then add to .env:
VIDEO_GEN_LOCAL_ENABLED=true
VIDEO_GEN_LOCAL_MODEL=wan2.1-1.3bPipelines Overview
OpenMontage includes 12 production pipelines, each a complete workflow from idea to finished video:
| Pipeline | What It Produces | Best For |
|---|---|---|
| Animated Explainer | AI-generated explainer with research, narration, music | Educational content, tutorials |
| Animation | Motion graphics, kinetic typography | Social media, product demos |
| Avatar Spokesperson | Avatar-driven presenter videos | Corporate comms, training |
| Cinematic | Trailer, teaser, mood-driven edits | Brand films, promotional content |
| Clip Factory | Batch of ranked short-form clips from one source | Repurposing long content |
| Documentary Montage | Thematic montage from free stock footage | Video essays, mood pieces |
| Hybrid | Source footage + AI-generated support visuals | Enhancing existing footage |
| Localization & Dub | Subtitle, dub, and translate existing video | Multi-language distribution |
| Podcast Repurpose | Podcast highlights to video | Podcast marketing |
| Screen Demo | Polished screen recordings and walkthroughs | Product demos, tutorials |
| Talking Head | Footage-led speaker videos | Presentations, vlogs |
Architecture
Agent-First Orchestration
OpenMontage uses an agent-first architecture. There is no Python orchestrator. Your AI coding assistant IS the orchestrator:
- Agent reads the pipeline manifest (YAML) — defines stages, tools, review criteria, success gates
- Agent reads stage director skills (Markdown) — teaches the agent exactly how to execute each stage
- Agent calls Python tools — the tool registry auto-discovers 52+ tool implementations
- Agent self-reviews using reviewer skills — schema validation, playbook compliance, quality checks
- Agent checkpoints state (JSON) — resumable, with decision log and cost snapshot
- Pre-compose validation gate — delivery promise, slideshow risk, renderer governance
- Render — Remotion (React-based) or FFmpeg composition
- Post-render self-review — ffprobe, frame extraction, audio analysis, promise verification
Three-Layer Knowledge Architecture
Layer 1: tools/ + pipeline_defs/ "What exists" — executable capabilities + orchestration
Layer 2: skills/ "How to use it" — OpenMontage conventions and quality bars
Layer 3: .agents/skills/ "How it works" — external technology knowledge packsEach tool declares which Layer 3 skills it relies on. The agent reads Layer 1 to know what's available, Layer 2 to know how OpenMontage wants it used, and Layer 3 for deep technical knowledge.
Tool System
All 52+ tools inherit from BaseTool and declare: name, version, tier, capability, provider, runtime, stability, dependencies, input/output schemas, fallback chain, agent skills, and resource profile.
The ToolRegistry singleton auto-discovers all tools — no manual registration needed. Key queries include get_by_capability("tts"), get_available(), support_envelope(), and find_fallback().
Scored Provider Selection
Every tool selection runs through a 7-dimension scoring engine:
- Task fit (30%) — how well the provider matches the specific task
- Output quality (20%) — expected quality of generated assets
- Control features (15%) — parameterization and customization
- Reliability (15%) — uptime and consistency
- Cost efficiency (10%) — value for money
- Latency (5%) — speed of generation
- Continuity (5%) — provider stability
The winning provider and its score are logged in the decision trail with all alternatives considered.
Production Governance
OpenMontage treats video production like real engineering:
- Pre-compose validation — blocks render if delivery promise is violated, slideshow risk score is critical, or renderer family is missing
- Post-render self-review — ffprobe validation, frame extraction at 4 positions, audio level analysis, delivery promise verification
- Slideshow risk scoring — 6-dimension analysis prevents "animated PowerPoint" outputs
- Source media inspection — probes every file (resolution, codec, audio channels, duration)
- Decision audit trail — every creative and technical choice logged with alternatives and reasoning
- Budget controls — estimate before execution, reserve funds, reconcile after, configurable modes (observe/warn/cap)
Style System
Three built-in visual style playbooks control typography, color palettes, motion styles, and audio profiles:
| Playbook | Best For |
|---|---|
| Clean Professional | Corporate, educational, SaaS |
| Flat Motion Graphics | Social media, TikTok, startups |
| Minimalist Diagram | Technical deep-dives, architecture |
Platform Output Profiles
Built-in render profiles for every major platform: YouTube Landscape (1920×1080), YouTube 4K (3840×2160), YouTube Shorts (1080×1920), Instagram Reels (1080×1920), Instagram Feed (1080×1080), TikTok (1080×1920), LinkedIn (1920×1080), and Cinematic (2560×1080, 21:9).
What You Get With Zero API Keys
Out of the box, make setup gives you:
- Narration — Piper TTS, free offline text-to-speech
- Open footage — Archive.org + NASA + Wikimedia Commons
- Extra stock — Pexels + Unsplash + Pixabay (developer keys are free)
- Composition — Remotion (React-based) and HyperFrames (HTML/GSAP)
- Post-production — FFmpeg encoding, subtitle burn-in, audio mixing, color grading
- Subtitles — Auto-generated captions with word-level timing
Verification Checklist
make setupcompletes without errors- The agent reads pipeline manifests and stage director skills
- Research stage runs live web searches
- Script and scene plan are generated
- Assets (images, audio) are created
- Video renders successfully via Remotion or FFmpeg
- Post-render self-review passes (ffprobe, frame sampling, audio analysis)
- Final video is playable and contains correct narration + visuals
Resources
- GitHub: github.com/calesthio/OpenMontage
- Documentation: github.com/calesthio/OpenMontage#readme
- Architecture: github.com/calesthio/OpenMontage/blob/main/docs/ARCHITECTURE.md
- Provider Guide: github.com/calesthio/OpenMontage/blob/main/docs/PROVIDERS.md
- Prompt Gallery: github.com/calesthio/OpenMontage/blob/main/PROMPT_GALLERY.md
- YouTube Channel: youtube.com/@OpenMontage
- Community: GitHub Discussions
- License: GNU AGPLv3