- Python 3.11 or higher
- pip (Python package installer)
- API keys for your chosen providers
- Internet connection (for web search grounding)
pip install filepromptforgegit clone https://github.com/morganross/FilePromptForge-v2.git
cd FilePromptForge-Standalone
pip install -e .Create a .env file in the FilePromptForge/ directory:
# OpenAI (required for OpenAI models)
OPENAI_API_KEY=sk-...
# Google Gemini
GOOGLE_API_KEY=...
# Anthropic (via OpenRouter)
ANTHROPIC_API_KEY=...
# OpenRouter
OPENROUTER_API_KEY=...
# Perplexity
PERPLEXITY_API_KEY=...
# Tavily
TAVILY_API_KEY=...Note: The .env file is the canonical source for API keys. FilePromptForge will NOT fall back to mock responses if keys are missing - it will fail with a clear error.
# Test CLI
fpf --help
# Test Python import
python -c "from FilePromptForge import file_handler; print('OK')"# Create test files
echo "This is a test document." > test_input.txt
echo "Summarize this document." > test_instructions.txt
# Run FPF
fpf --file-a test_input.txt --file-b test_instructions.txt \
--out result.md --provider openai --model gpt-5-mini- Ensure
.envfile exists inFilePromptForge/directory - Check that the key name matches (e.g.,
OPENAI_API_KEYnotOPENAI_KEY) - Verify the key is valid by testing it directly with the provider
- Check that the model is in the provider's allowed list
- For OpenAI: gpt-5, gpt-5-mini, o3, o4-mini are supported
- For Google: gemini-2.5-pro, gemini-3-pro-preview are supported
- This is intentional - FPF requires provider-side web search
- Ensure your model supports web search (most recent models do)
- Check the provider's documentation for web search support