Overview

Fastfold agent integrations let you access folding, workflows, and related research operations through conversational or agent-driven interfaces instead of calling the SDK or raw API directly. Use them when you want an agent to interpret requests, coordinate multi-step tasks, and surface Fastfold capabilities inside the tools you already use.

This section covers the main ways to work with Fastfold through agents:

• Agent CLI for terminal-first natural-language research workflows
• Skills for reusable Fastfold capabilities inside supported coding agents
• PyMOL Agent for chat-driven structure workflows inside PyMOL

How to choose

• Use Agent CLI when you want one agent to interpret a request, choose tools, and execute a multi-step workflow from the terminal.
• Use Skills when you already work in an agent environment such as Cursor, Claude Code, or Codex and want Fastfold operations exposed there as packaged capabilities.
• Use PyMOL Agent when structure inspection and visualization inside PyMOL are part of the workflow.

What each path is best at

Agent CLI

Best for:

• interactive research sessions in the terminal
• autonomous multi-step analysis
• background task workflows for long-running fold or workflow jobs
• report export and Slack sharing from the same session

Skills

Best for:

• adding Fastfold workflows to external agent tools
• standardized create -> wait -> results patterns
• reusable agent automation for folding, MD, BoltzGen, and report sharing
• teams that want the same Fastfold operations available across multiple agent environments

PyMOL Agent

Best for:

• combining agent guidance with in-app structure visualization
• loading fold or workflow outputs directly into PyMOL
• structure-focused workflows that benefit from interactive inspection

Relationship to SDK and API

Agents complement the direct integration paths:

• use /sdk when you want explicit Python or SDK CLI control
• use /api when you want raw HTTP integration
• use the agent paths in this section when you want higher-level orchestration or a conversational workflow

What's next

• New to Fastfold? Follow /quick-start to create an API key and run a first fold or workflow, then pick an agent path above.
• Installing the Agent CLI: prerequisites, uv tool install, auth, and commands → /agents/cli.
• Adding Skills to Cursor, Claude Code, or Codex (install, env vars, prompts) → /agents/skills.
• PyMOL Agent open-source vs commercial install, first run, and troubleshooting → /agents/pymol.
• Prefer Python or HTTP? Use the SDK or API reference; agents sit on top of the same jobs and workflows.