openmontage

openclaw-openmontage

An OpenClaw plugin that adds AI video production capabilities via a local OpenMontage workspace.

What it does

Injects skills that teach OpenClaw agents to delegate video production tasks — animated explainers, cinematic trailers, avatar videos, screen demos, motion graphics, and more — to your local OpenMontage installation.

The plugin contributes no runtime JS code. It is a skills package: structured markdown files that get injected into the agent's system prompt and teach it how to engage OpenMontage's instruction-driven pipeline system.

Requirements

• OpenClaw ≥ 2026.1.26
• OpenMontage cloned locally and set up

(run make setup in the OpenMontage repo)

Install

openclaw install @itsuzef/openclaw-openmontage

Or from a local checkout:

{
  "plugins": {
    "installs": [
      { "source": "path", "sourcePath": "/path/to/openclaw-openmontage" }
    ]
  }
}

Configure

openclaw config set plugins.entries.openmontage.enabled true
openclaw config set plugins.entries.openmontage.config.workspacePath /path/to/OpenMontage

Or in your openclaw config JSON directly:

{
  "plugins": {
    "entries": {
      "openmontage": {
        "enabled": true,
        "config": {
          "workspacePath": "/Users/you/Documents/OpenMontage"
        }
      }
    }
  }
}

Usage

Once configured, ask your OpenClaw agent anything video-related:

• "Make a 60-second animated explainer about how black holes form"
• "Create a Ghibli-style animated video of an underwater city at dusk"
• "Build a cinematic trailer for our product launch"
• "Make a screen demo showing how to install our CLI tool"
• "Turn this blog post into a 90-second video"
• "25-second kinetic typography launch reel using HyperFrames"

The agent will run preflight, present a production plan and cost estimate, and execute the pipeline with human approval gates at each creative stage.

Cost

Most pipelines have a zero-cost path using free tools (Piper TTS + stock media + Remotion + FFmpeg). Premium paths add AI image/video generation and ElevenLabs TTS.

| Path | Cost | Unlocks | |---|---|---| | Zero key | $0 | Pipelines with a free path (explainer, animation, screen-demo, hybrid, clip-factory) | | FAL_KEY | ~$0.15–$1.50 | FLUX AI image generation | | Full setup | ~$1–$3 | Video gen (Veo/Kling/Runway/Seedance) + ElevenLabs + music |

Pipelines

| Pipeline | Best For | Status | |---|---|---| | animated-explainer | Explainers, educational content, data viz | production | | cinematic | Trailers, teasers, hype edits | production | | animation | Motion graphics, Ghibli/anime-style | production | | screen-demo | CLI demos, developer walkthroughs | production | | avatar-spokesperson | Presenter videos, announcements | production | | hybrid | Source footage + AI support visuals | production | | talking-head | Speaker videos | beta | | clip-factory | Multi-clip from long source | beta | | podcast-repurpose | Podcast highlights | beta | | localization-dub | Translate and dub | beta | | documentary-montage | Real-footage montage, tone-poem style | beta |

How it works

OpenMontage is agent-driven: the AI agent reads pipeline manifests and stage director skills, makes creative decisions, and uses Python tools for actual generation. This plugin teaches OpenClaw agents:

1. What OpenMontage can produce (skills/openmontage/openmontage.md) 2. How to engage the workspace — preflight, pipeline selection, plan presentation (skills/openmontage/delegate.md) 3. Which pipeline fits which brief (skills/openmontage/pipelines.md)

The agent navigates to your OpenMontage workspace using its existing file and shell tools, reads AGENT_GUIDE.md from that workspace, and follows the instruction-driven pipeline system from there.

Contributing

Improvements to the skill content are the most valuable contributions. If OpenMontage adds a new pipeline or changes its capability surface, update the matching skill file.

1. Fork and clone this repo 2. Edit skills/openmontage/*.md 3. Test by pointing your local plugin install at the fork 4. Open a PR with a clear description of what changed and why