JSPM – pixelmuse@0.4.1

Package Exports

This package does not declare an exports field, so the exports above have been automatically detected and optimized by JSPM instead. If any package subpath is missing, it is recommended to post an issue to the original package (pixelmuse) to support the "exports" field. If that is not possible, create a JSPM override to customize the exports field for this package.

Readme

AI image generation from the terminal.
CLI, interactive TUI, and MCP server — powered by the Pixelmuse API.

Pixelmuse CLI demo

Why Pixelmuse?

One command, any model. Flux, Imagen 3, Recraft V4, and more — switch models with a flag, no separate accounts or API keys.
Built for developer workflows. Pipe from stdin, JSON output for scripting, watch mode for prompt iteration, MCP server for AI agents.
Predictable credit pricing. 1–7 credits per generation, most models just 1 credit. Free credits on signup.

Get Started

npm install -g pixelmuse
pixelmuse setup

The setup wizard creates your account (opens browser), configures your API key, and optionally sets up the MCP server for Claude Code, Cursor, or Windsurf. New accounts include 15 free credits.

Requires Node.js 20+. For terminal image previews, install chafa (brew install chafa / sudo apt install chafa).

Quick Start

# Generate an image
pixelmuse "a cat floating through space"

# Choose model and aspect ratio
pixelmuse "neon cityscape at night" -m recraft-v4 -a 16:9

# Apply a style
pixelmuse "mountain landscape" -s anime -a 21:9

# Save to specific path
pixelmuse "app icon, minimal" -o icon.png

Account and model overview

Prompt Templates

Save reusable prompts as YAML files with variables and default settings.

# Scaffold a new template
pixelmuse template init product-shot

# Generate with a template
pixelmuse template use blog-thumbnail --var subject="React hooks guide"

# List all templates
pixelmuse template list

Templates are stored at ~/.config/pixelmuse-cli/prompts/:

# blog-thumbnail.yaml
name: Blog Thumbnail
description: Dark-themed blog post thumbnail
prompt: >
  A cinematic {{subject}} on a dark gradient background,
  dramatic lighting, 8K resolution
defaults:
  model: nano-banana-2
  aspect_ratio: "16:9"
variables:
  subject: "code editor with syntax highlighting"
tags: [blog, thumbnail, dark]

Interactive TUI

A full terminal UI for visual browsing, generation wizards, gallery, and account management:

pixelmuse ui

Interactive TUI generation

Scripting & Pipes

# Pipe prompt from stdin
echo "hero banner for SaaS landing page" | pixelmuse -o hero.png
cat prompt.txt | pixelmuse -m recraft-v4

# JSON output for scripting
pixelmuse --json "logo concept" | jq .output_path

# Watch mode — regenerates when prompt file changes
pixelmuse --watch prompt.txt -o output.png

# Skip preview, copy to clipboard
pixelmuse "avatar" --no-preview --clipboard

CLI help and commands

MCP Server (Claude Code, Cursor, Windsurf)

The MCP server lets AI agents generate images, list models, and check your balance — no manual CLI steps needed.

Claude Code — add to ~/.claude/.mcp.json:

{
  "mcpServers": {
    "pixelmuse": {
      "command": "npx",
      "args": ["-y", "pixelmuse-mcp"],
      "env": {
        "PIXELMUSE_API_KEY": "pm_live_your_key_here"
      }
    }
  }
}

Cursor / Windsurf — add to your MCP settings:

{
  "pixelmuse": {
    "command": "npx",
    "args": ["-y", "pixelmuse-mcp"],
    "env": {
      "PIXELMUSE_API_KEY": "pm_live_your_key_here"
    }
  }
}

The agent gets three tools:

Tool	What it does
`generate_image`	Generate an image from a prompt with model, aspect ratio, style, and output path.
`list_models`	List all available models with credit costs.
`check_balance`	Check your credit balance and plan info.

Example prompts for your AI agent:

"Generate a hero image for my landing page, 16:9, save to ./public/hero.png"
"Create a blog thumbnail about React hooks using the anime style"
"What Pixelmuse models are available?"

Which Interface?

Pixelmuse ships four interfaces. Pick the one that fits your workflow — they all use the same API and credentials.

	CLI	TUI	MCP Server	Claude Skill
What it is	Command-line tool	Interactive terminal UI	AI agent tool server	Claude Code prompt template
Launch	`pixelmuse "prompt"`	`pixelmuse ui`	Auto-starts with Claude/Cursor	Auto-triggers on keywords
Best for	Scripting, automation, CI/CD	Visual browsing, exploring models	Letting AI agents generate images	Generating images mid-conversation
Input	Flags, stdin, pipe, watch mode	Guided wizard with menus	AI decides params from natural language	Natural language to Claude

Rule of thumb:

You type the prompt → CLI or TUI
AI types the prompt → MCP Server or Claude Skill
Quick one-off → CLI
Browsing/exploring → TUI

Models

Model	Credits	Best For
Nano Banana 2 (default)	1	Speed, text rendering, world knowledge
Nano Banana Pro	4	Text rendering, real-time info, multi-image editing
Flux Schnell	1	Quick mockups, ideation
Google Imagen 3	1	Realistic photos, complex compositions
Recraft V4	1	Typography, design, composition
Recraft V4 Pro	7	High-res design, art direction

CLI Reference

Commands

Command	Description
`pixelmuse setup`	First-time setup wizard (account, MCP, defaults)
`pixelmuse "prompt"`	Generate an image (default command)
`pixelmuse models`	List available models with costs
`pixelmuse account`	Account balance and usage stats
`pixelmuse history`	Recent generations table
`pixelmuse open <id>`	Open a generation in system viewer
`pixelmuse login`	Authenticate with API key
`pixelmuse logout`	Remove stored credentials
`pixelmuse template <cmd>`	Manage prompt templates
`pixelmuse ui`	Launch interactive TUI

Flags

Flag	Description
`-m, --model`	Model ID (default: `nano-banana-2`)
`-a, --aspect-ratio`	Aspect ratio (default: `1:1`)
`-s, --style`	`realistic`, `anime`, `artistic`, `none`
`-o, --output`	Output file path
`--json`	Machine-readable JSON output
`--no-preview`	Skip terminal image preview
`--open`	Open result in system viewer
`--clipboard`	Copy image to clipboard
`--watch <file>`	Watch prompt file, regenerate on save
`--no-save`	Don't save image to disk

Configuration

Settings at ~/.config/pixelmuse-cli/config.yaml:

defaultModel: nano-banana-2
defaultAspectRatio: "1:1"
defaultStyle: none
autoPreview: true
autoSave: true

Path	Contents
`~/.config/pixelmuse-cli/config.yaml`	User settings
`~/.config/pixelmuse-cli/auth.json`	API key (fallback if keychain unavailable)
`~/.config/pixelmuse-cli/prompts/`	Prompt template YAML files
`~/.local/share/pixelmuse-cli/generations/`	Auto-saved generation images

License

Business Source License 1.1 (BSL 1.1). See LICENSE for details.

Free for any use except offering a competing image generation API or platform. Converts to GPL-2.0 on 2029-03-01.

pixelmuse