Quick Start

Use this page to get to a first successful run quickly, then move into the deeper product-specific docs.

Choose an integration path

• Python SDK and CLI for scripts, notebooks, CI, payload files, YAML submit, and workflow helpers
• Agents for Agent CLI, Skills, and PyMOL Agent
• REST API for direct HTTP integrations

Python SDK and CLI

Use this path when you want programmatic control over fold jobs, workflows, uploads, and reports.

1. Create an API key

Store the key in FASTFOLD_API_KEY so both the SDK and CLI pick it up automatically.

export FASTFOLD_API_KEY="sk-...your-api-key"

2. Install the package

Requires Python 3.8+.

pip install fastfold-ai

3. Run a first fold in Python

from fastfold import Client

client = Client()

job = client.fold.create(
    sequence="LLGDFFRKSKEKIGKEFKRIVQRIKDFLRNLVPRTES",
    model="boltz-2",
    is_public=True,
)

results = client.jobs.wait_for_completion(job.id, poll_interval=5.0, timeout=900.0)
print("Status:", results.job.status)
print("CIF URL:", results.cif_url())
print("Mean PLDDT:", results.metrics().mean_PLDDT)
print("Viewer:", results.get_viewer_link())

4. Run the same fold from the SDK CLI

fastfold-cli fold --sequence "LLGDFFRKSKEKIGKEFKRIVQRIKDFLRNLVPRTES" --model boltz-2

On success, the SDK CLI prints the created job ID to stdout.

5. Move to file-backed and workflow-driven runs

Submit a full job payload file:

fastfold-cli jobs create --payload-file fastfold/examples/fold/job_payload.json

Submit Boltz-style YAML directly:

fastfold-cli jobs from-yaml --file fastfold/examples/fold/boltz2_affinity_input.yaml --model boltz-2

Start an OpenMM workflow from local structure files:

fastfold-cli workflows openmm from-manual-files --pdb ./protein.pdb --pae ./pae.json --simulation-name AF-P00698 --sim-length-ns 10 --box-length 50

Start an OpenMMDL workflow from local topology and ligand files:

fastfold-cli workflows openmmdl from-local-files --topology ./KEAP1kd.pdb --ligand ./IQK.sdf --simulation-name "KEAP1 + IQK" --input-json fastfold/examples/openmmdl/workflow_input.json

Use /sdk for the full recipe set, including OpenMM, OpenMMDL, BoltzGen, generic workflows, webhooks, and report sharing.

Agent CLI

Use the Agent CLI when you want Fastfold to interpret a request, choose tools, and run a multi-step workflow for you.

Install

uv tool install "fastfold-agent-cli[all]" --python 3.10

Authenticate

fastfold setup

This interactive setup walks through the required credentials.

Run a first query

fastfold "Fold this sequence with boltz-2 and show me the binding pockets: MALWMRLLPLL..."

fastfold

Install, setup, and full command reference → Agent CLI

PyMOL Agent

Use Fastfold from inside PyMOL with a plugin-based chat workflow. There are two install paths—pick the one that matches how you run PyMOL.

Open Source install

Use this when you do not already have PyMOL, or you want the Fastfold-managed open source stack (conda + installer script).

curl -LsSf https://fastfold.ai/pymol-agent/install.sh | sh

If PyMOL is already in your environment and you only need the agent:

curl -LsSf https://fastfold.ai/pymol-agent/install.sh | sh -s -- --agent-only

Commercial PyMOL install

If you use commercial PyMOL and prefer Plugin Manager installs, use the plugin zip from GitHub Releases instead of the curl installer:

https://github.com/fastfold-ai/fastfold-pymol-agent/releases/latest/download/fastfold-pymol-agent-plugin.zip

Install via Plugin > Plugin Manager > Install New Plugin, then run fastfold deps install from the PyMOL command line once. Full steps: PyMOL Agent.

Docs and repo

• Guide: PyMOL Agent (open source vs commercial, first run, troubleshooting)
• Repository: fastfold-ai/fastfold-pymol-agent

Next steps

• SDK install: /sdk/install
• SDK overview and recipes: /sdk
• SDK CLI recipes: /sdk/cli
• Agents overview: /agents
• Agent CLI: /agents/cli
• PyMOL Agent: /agents/pymol
• API Reference: /api
• Available models: cloud.fastfold.ai/models