Install¶

Python 3.11+ is required.

Default: extraction only¶

pip install .
# or published package:
pip install figma-extractor
conda install -c vickykumar7125 figma-extractor

From a requirements file:

pip install -r requirements.txt

That file is the extraction list, with no version pins. Development tests:

pip install -e ".[dev]"

extract and info work after this install. They do not import LangChain or PyTorch.

Maximum features for this machine¶

From the repository root:

python setup.py

The installer prints the detected OS, CPU architecture, accelerator, and the pip commands it will run. See the plan without installing:

python setup.py --dry-run

Flag	Effect
(none)	Editable package, then `cuda132.txt`, `cuda130.txt`, `cuda129.txt`, `cpu.txt`, `xpu.txt`, `gpu.txt`, or `macos.txt`
`--core`	Extraction only, same result as `pip install -e .`
`--no-llm`	Skip LangChain and provider packages
`--no-torch`	Skip PyTorch and torchvision
`--dry-run`	Print the plan and commands

Detection order:

Environment	Profile	Requirements file
macOS Apple Silicon	MPS via the PyPI wheel	`requirements/macos.txt`
Linux or Windows, driver CUDA 13.2 or newer	torch `2.14.1+cu132`, torchvision `0.29.1+cu132`	`requirements/cuda132.txt`
Linux or Windows, driver CUDA 13.0 or 13.1	torch `2.14.1+cu130`, torchvision `0.29.1+cu130`	`requirements/cuda130.txt`
Linux or Windows, driver CUDA 12.9	Linux: torch `2.13.0+cu129` / torchvision `0.28.0+cu129`. Windows: torch `2.8.0+cu129` / torchvision `0.23.0+cu129`	`requirements/cuda129.txt`
Linux with ROCm and no NVIDIA GPU	ROCm index `rocm7.1`	`requirements/gpu.txt`
Linux or Windows with `xpu-smi` or `sycl-ls`	Intel XPU index	`requirements/xpu.txt`
Linux or Windows otherwise	CPU index	`requirements/cpu.txt`

python setup.py reads the CUDA version from nvidia-smi and chooses one of the three CUDA files. A driver newer than 13.2 uses cuda132.txt. Those three files pin torch and torchvision to the pair published on that index, checked on 2026-10-02. CPU, XPU, GPU, and macOS files have no version pins. macOS Intel is reported and the profile is skipped. The editable package is installed first, then the selected file.

Install one feature yourself¶

pip install -e ".[openai]"          # also: anthropic, google, vertex, ollama,
                                    # huggingface, groq, xai, nvidia, cohere,
                                    # together, deepseek
pip install -e ".[llm]"             # LangGraph runtime, no chat provider
pip install -e ".[all-llm]"         # every chat provider, no Transformers or torch
pip install -e ".[huggingface-local]"
pip install -e ".[huggingface-quant]"   # NVIDIA CUDA bitsandbytes
pip install -r requirements/cuda132.txt
pip install -r requirements/cuda130.txt
pip install -r requirements/cuda129.txt
pip install -r requirements/cpu.txt
pip install -r requirements/xpu.txt
pip install -r requirements/gpu.txt

requirements/common.txt is the shared chat-provider list. The accelerator files include it.

Credentials stay in the environment. They are not accepted in JSON config files.

Provider id	Extra	Environment variable
`openai`	`openai`	`OPENAI_API_KEY`
`anthropic`	`anthropic`	`ANTHROPIC_API_KEY`
`google`	`google`	`GOOGLE_API_KEY`
`vertex`	`vertex`	`GOOGLE_CLOUD_PROJECT` and Application Default Credentials
`anthropic-vertex`	`vertex`	`GOOGLE_CLOUD_PROJECT` and Application Default Credentials
`ollama`	`ollama`	none (`OLLAMA_BASE_URL` optional)
`huggingface`	`huggingface`	`HF_TOKEN` or `HUGGINGFACEHUB_API_TOKEN` only for `HF_BACKEND=remote`
`groq`	`groq`	`GROQ_API_KEY`
`xai`	`xai`	`XAI_API_KEY`
`nvidia`	`nvidia`	`NVIDIA_API_KEY`
`cohere`	`cohere`	`COHERE_API_KEY`
`together`	`together`	`TOGETHER_API_KEY`
`deepseek`	`deepseek`	`DEEPSEEK_API_KEY`

anthropic calls the Anthropic API. anthropic-vertex calls Claude on Vertex AI. Local Hugging Face execution is the default for that provider. cuda132.txt, cuda130.txt, and cuda129.txt include bitsandbytes. The other profiles do not. See Hugging Face local.

Default model when --llm-model and LLM_MODEL are omitted:

Provider id	Default model
`openai`	`gpt-4.1-mini`
`anthropic`	`claude-sonnet-4-5`
`google`	`gemini-2.5-flash`
`vertex`	`gemini-2.5-flash`
`anthropic-vertex`	`claude-haiku-4-5@20251001`
`ollama`	`llama3.2`
`huggingface`	local: `HF_LOCAL_MODEL_PATH`. Remote: `microsoft/Phi-3-mini-4k-instruct`
`groq`	`llama-3.3-70b-versatile`
`xai`	`grok-3`
`nvidia`	`meta/llama-3.1-70b-instruct`
`cohere`	`command-r-plus`
`together`	`meta-llama/Llama-3.3-70B-Instruct-Turbo`
`deepseek`	`deepseek-chat`

After torch is installed:

figma-extractor devices

That reports cpu, cuda, rocm, mps, xpu, or unavailable.