Brandly CLI

AI product video orchestrator — add image, video, and sound generation capability to any AI tool (OpenCode, Codex, Qwen Code, Claude Code, etc.) via a CLI and Director agent.

What is Brandly?

Brandly is an autonomous video production pipeline that turns product ideas into platform-ready marketing videos. It provides:

• Image generation via Agnes AI (text-to-image, image-to-image)
• Video generation via Agnes AI (text-to-video, keyframe-controlled, reference-based)
• Audio generation via MiniMax Audio (background music, TTS voiceover)
• Multi-agent pipeline — automated workflow from idea → trends → concept → script → assets → audio → validate → publish
• Director orchestrator — an autonomous agent that guides the entire production process
• Credit budgeting — track spend per phase against a project budget
• Style presets — avoid AI slop with photorealistic, cinematic, editorial, commercial, and documentary presets

v0.3.0 New Features

Installation

pip install brandly-cli

Configure API Keys

# Agnes AI (image + video generation)
export AGNES_API_KEY="your-agnes-api-key"

# MiniMax Audio (music + TTS)
export MINIMAX_API_KEY="your-minimax-api-key"

Or create a .env file in your project root:

AGNES_API_KEY=your_key_here
MINIMAX_API_KEY=your_key_here
AGNES_BASE_URL=https://apihub.agnes-ai.com/v1
MINIMAX_BASE_URL=https://api.minimaxi.com/v1

Quick Start

1. Initialize a Project

brandly init \
  --name "SuperWidget Pro" \
  --idea "A revolutionary widget that organizes your desk with AI" \
  --style cinematic \
  --budget 500 \
  --shots 5 \
  --platforms tiktok instagram youtube

2. Run the Pipeline

# Check cost estimate first
brandly estimate --style cinematic --shots 5

# Run each phase
brandly run <project-id>        # Start the current phase
brandly approve <project-id> <phase>  # Approve and advance

# Or run the full pipeline at once
brandly director                # Shows Director prompt for AI tools

3. Generate Media Directly

# Image generation
brandly image --prompt "product on marble surface, studio lighting" \
               --style-preset cinematic

# Video generation
brandly video <project-id> --prompt "sleek wireless earbuds rotating" \
             --duration 5 --wait

# Audio generation
brandly music --prompt "upbeat electronic background music" --duration 30
brandly tts "Welcome to SuperWidget Pro"

4. Export & Share

# Export for platforms
brandly export <project-id> --platforms tiktok youtube

# Generate thumbnails
brandly thumbnail <project-id>

# Stitch clips together
brandly stitch clip1.mp4 clip2.mp4 --transition fade

# Dub to another language
brandly voice-match input.mp4 --source en --target es

# Share to cloud
brandly share output.mp4 --provider local

CLI Reference

Project Management

Media Generation

Post-Production

Intelligence

System

Pipeline Phases

init → trends → concept → script → asset → audio → re_edit → validate → publish → done

Each phase has a corresponding agent prompt that defines the task for the AI orchestrator:

Video Styles

Style Presets (Avoid AI Slop)

Director Mode

The Director is an autonomous orchestrator agent. When invoked by an AI tool, it:

1. Gathers product info (name, idea, style, budget, platforms)
2. Creates a project via brandly init
3. Runs the pipeline phase by phase
4. Checks virality, manages budget, and exports final assets

To use it with any AI tool, copy the output of:

brandly director

Project Structure

.brandly/
├── projects/
│   └── {project-id}/
│       ├── project.json      # Project state
│       ├── cost.json         # Credit spend log
│       └── artifacts/
│           └── {phase}/      # Phase outputs (.md, .json)
├── user-preferences.json     # Liked/disliked hooks, style prefs
└── templates/                # Custom saved templates

Generated media is stored in:

• imagen/{project-id}/ — generated images
• videgen/{project-id}/ — generated videos
• audgen/{project-id}/ — generated audio

Development

# Install in development mode
pip install -e ".[dev]"

# Run tests
pytest

# Run tests with coverage
pytest --cov=src/brandly_cli

# Lint
ruff check src/
ruff check src/ --fix

# Syntax check
python -m py_compile src/brandly_cli/*.py

# Build
python -m build

Integration with AI Tools

OpenCode

Add to .opencode/package.json:

{ "dependencies": { "brandly": "github:Dream-Pixels-Forge/brandly-plugin" } }

Register in opencode.json:

{ "$schema": "https://opencode.ai/config.json", "plugin": ["brandly"] }

Codex / Qwen Code / Claude Code

Use the CLI directly — all commands are designed to be called by agent subprocesses. The Director prompt provides the orchestration logic.

License

MIT — Dream Pixels Forge