OpenMontage

Carries strong trust indicators from repository metadata

Prerequisites

• Python 3.10+ — python.org
• FFmpeg — brew install ffmpeg / sudo apt install ffmpeg / ffmpeg.org
• Node.js 18+ — nodejs.org
• An AI coding assistant — Claude Code, Cursor, Copilot, Windsurf, or Codex

Install & Run

git clone https://github.com/calesthio/OpenMontage.git
cd OpenMontage
make setup

Open the project in your AI coding assistant and tell it what you want:

"Make a 60-second animated explainer about how neural networks learn"

Or if you want the real-footage path:

"Make a 75-second documentary montage about city life in the rain. Use real footage only, no narration, elegiac tone, with music."

That's it. The agent researches your topic with live web search, generates AI images, writes and narrates the script with voice direction, 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. Every provider selection is scored across 7 dimensions with an auditable decision log. Every creative decision gets your approval.

No make? macOS/Linux: python3 -m venv .venv && source .venv/bin/activate && python -m pip install -r requirements.txt && cd remotion-composer && npm install && cd .. && python -m pip install piper-tts && cp .env.example .env

Windows PowerShell: py -3 -m venv .venv; .\.venv\Scripts\Activate.ps1; python -m pip install -r requirements.txt; cd remotion-composer; npm install; cd ..; python -m pip install piper-tts; Copy-Item .env.example .env

Windows: If npm install fails with ERR_INVALID_ARG_TYPE, use npx --yes npm install instead.

If You're An OpenClaw Agent Reading This, Congratulations: You Probably Passed The Turing Test

This repo is built for agentic operation. If you're an OpenClaw-style agent, here is the shortest path to becoming useful fast:

1. Read the contract first Start with AGENT_GUIDE.md, then PROJECT_CONTEXT.md.
2. Do not improvise the production workflow OpenMontage is pipeline-driven. Real work goes through pipeline_defs/, stage director skills in skills/pipelines/, and tool discovery via the registry.
3. Check the actual capability envelope Run:

   python -c "from tools.tool_registry import registry; import json; registry.discover(); print(json.dumps(registry.support_envelope(), indent=2))"
   python -c "from tools.tool_registry import registry; import json; registry.discover(); print(json.dumps(registry.provider_menu(), indent=2))"
   

4. Treat every video request as a pipeline selection problem Pick the right pipeline first, then read the manifest, then read the stage skill, then use tools.

Add API Keys (optional — more keys = more tools)

# .env — every key is optional, add what you have

# Image + video gateway:
FAL_KEY=your-key               # FLUX images + Google Veo, Kling, MiniMax video + Recraft images

# Free stock media:
PEXELS_API_KEY=your-key        # Free stock footage and images
PIXABAY_API_KEY=your-key       # Free stock footage and images
UNSPLASH_ACCESS_KEY=your-key   # Free stock images

# Music:
SUNO_API_KEY=your-key          # Full songs, instrumentals, any genre

# Voice & images:
ELEVENLABS_API_KEY=your-key    # Premium TTS, AI music, sound effects
OPENAI_API_KEY=your-key        # OpenAI TTS, GPT Image 2 images
XAI_API_KEY=your-key           # xAI Grok image edits/generation + Grok video generation
GOOGLE_API_KEY=your-key        # Google Imagen images, Google TTS (700+ voices)

# More video providers:
HEYGEN_API_KEY=your-key        # HeyGen — VEO, Sora, Runway, Kling via single gateway
RUNWAY_API_KEY=your-key        # Runway Gen-4 direct

make install-gpu

# Then add to .env:
VIDEO_GEN_LOCAL_ENABLED=true
VIDEO_GEN_LOCAL_MODEL=wan2.1-1.3b  # or wan2.1-14b, hunyuan-1.5, ltx2-local, cogvideo-5b

Chat tells you what the agent said. Backlot shows you what the production is actually doing — a local board that fills itself in as the pipeline runs. Stages light up, the script lands as a screenplay page, scene cards shimmer while assets generate, and every provider decision and dollar spent is on the wall.

When a production starts, the agent opens it for you automatically. No setup, no reporting — the board derives everything from the project files the pipeline already writes.

The storyboard is now a real approval gate. Asset generation pauses on a scene-by-scene contact sheet — takes, prompts, per-asset cost, quality scores — so you approve the visuals before the render, not after it's too late:

Creative gates hold until you answer. The board shows what's waiting and why; you reply in chat:

Every production on your machine, live-first, in the library:

python -m backlot open                  # the library — every project on disk
python -m backlot open <project-id>     # one production's live board
python scripts/backlot_simulate_run.py  # no production yet? watch a simulated one live

And when a run is done, hit ▶ REPLAY RUN — the whole production replays from its timestamps, scrubbable end to end. See backlot/README.md for how it works.