Installation

Installation

Requirements

• Python 3.10 or higher
• A Google Gemini API key (free from Google AI Studio)

Option 1: pip install (recommended)

pip install paperbanana

This gives you the CLI (paperbanana generate, paperbanana plot, paperbanana evaluate, paperbanana setup) and the Python API.

For MCP server support:

pip install paperbanana[mcp]

Option 2: uvx (MCP server only, no install)

If you only want the MCP server for use with Cursor, Claude Code, or another MCP client, uvx handles everything. No pip install, no virtual environment, no git clone.

Prerequisite: install uv:

curl -LsSf https://astral.sh/uv/install.sh | sh

Then configure your MCP client to run:

uvx --from "paperbanana[mcp]" paperbanana-mcp

See MCP Server Setup for client-specific configuration.

Option 3: From source (for development)

git clone https://github.com/llmsresearch/paperbanana.git
cd paperbanana
pip install -e ".[dev,google]"

This installs additional dependencies for testing (pytest) and linting (ruff).

First-Time Setup

Run the setup wizard to configure your API key:

paperbanana setup

This opens your browser to Google AI Studio, where you can create a free API key. The wizard saves it to a .env file in the project root.

Manual Setup

cp .env.example .env

Edit .env and add your key: GOOGLE_API_KEY=<your-key-here>

Verify Installation

paperbanana generate \
  --input examples/sample_inputs/transformer_method.txt \
  --caption "Overview of our encoder-decoder architecture with sparse routing"

If output appears in outputs/run_<timestamp>/final_output.png, everything is working.

Next Steps

• Quick Start Guide for your first diagram
• MCP Server Setup for IDE integration
• CLI Reference for all available commands