Skip to content

CLAUDE.md

Latest commit

 

History

History
83 lines (61 loc) · 3.39 KB

CLAUDE.md

File metadata and controls

83 lines (61 loc) · 3.39 KB

CLAUDE.md

This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.

Quick Start

# Build
dotnet build src/ImageGenCli.csproj

# Run directly (development)
dotnet run --project src/ImageGenCli.csproj -- "Your prompt here"

# Install as global tool (recommended)
dotnet pack src/ImageGenCli.csproj -c Release -o ./nupkg
dotnet tool install --global --add-source ./nupkg ImageGenCli

# Then use anywhere
image-gen "Your prompt here"

Architecture

Multi-provider image generation CLI using .NET 10 and System.CommandLine.

Provider Pattern:

  • IImageGenerationClient - common interface for all providers
  • GeminiImageClient - Google Gemini API (gemini-2.5-flash-image, gemini-3-pro-image-preview)
  • OpenAIImageClient - OpenAI API (gpt-image-2, gpt-image-1.5, gpt-image-1, gpt-image-1-mini)
  • BflImageClient - Black Forest Labs FLUX API (flux-2-pro, flux-2-flex, flux-2-max)
  • PoeImageClient - Poe.com API (unified access to many models)

Models (in Models/):

  • GenerationRequest - provider-agnostic input (prompt, reference images, aspect ratio, etc.)
  • GenerationResult - output with images array and optional text response
  • GeneratedImage - single image with mime type and byte data

Adding a new provider:

  1. Create NewProviderImageClient.cs implementing IImageGenerationClient
  2. Add provider option to Program.cs switch statements (client instantiation, API key lookup, validation)

Provider-Specific Constraints

Parameters validated at runtime - using unsupported options with a provider causes a hard error:

Parameter Gemini OpenAI BFL Poe
--system-prompt Supported Error Error Error
--temperature Supported (0.0-2.0) Error Error Error
--resolution Pro models only gpt-image-2 only Supported Model-dependent
--quality Error Supported Error Supported
--samples max 4 10 10 10
--images max N/A N/A 8 N/A

BFL Model Comparison:

  • flux-2-pro (default) - Fast (~10s), production-ready, $0.03/MP
  • flux-2-flex - Adjustable controls (steps, guidance), best typography, $0.06/MP
  • flux-2-max - Highest quality, web grounding, $0.07/MP

Poe Model Access: Poe provides unified access to many image models (OpenAI, FLUX, Imagen, etc.) via single API. Use image-gen --list-models -p poe to see available models. Model names are case-sensitive.

Environment Variables

  • GEMINI_API_KEY - for Gemini provider
  • OPENAI_API_KEY - for OpenAI provider
  • BFL_API_KEY - for BFL (FLUX) provider
  • POE_API_KEY - for Poe provider

Agent Skill

This project includes an Agent Skill definition in image-gen/SKILL.md for AI agent integration. The skill provides structured instructions for agents to use the CLI effectively.

Notes

  • .tmp/ directory is gitignored and can be used for test outputs
  • Gemini's imageSize parameter only works with Pro models; Flash models ignore it
  • OpenAI maps aspect ratios to fixed pixel dimensions (no resolution control)
  • BFL uses async API with polling; dimensions must be multiples of 16
  • BFL doesn't support batch generation; multiple samples are generated sequentially
  • Poe returns images as markdown URLs in response content; client parses and downloads
  • Use --list-models -p <provider> to see available models for each provider