Help center

How do I access Higgsfield via CLI?

Updated

CLI and Skills is how Higgsfield integrates with coding agents. The difference from MCP: MCP is a full connection for chat agents that exposes all tools with free parameter selection, while CLI and Skills is optimized for coding agents, where each skill has a defined structure that reduces token overhead and produces consistent output. Accessing Higgsfield through the CLI falls under the Developer Terms in the Terms of Use: keep your session secure, and you're responsible for what a coding agent generates through your account. Both use the same Higgsfield account and plan credits with no API key, and Unlimited models and free generations don't apply to either. Landing page: higgsfield.ai/skills.

When should I use CLI instead of MCP?

When should I use CLI instead of MCP?

Criteria

CLI + Skills

MCP

Best for

Coding agents (Claude Code, Cursor, Codex)

Chat agents (Claude web, Cowork)

Token overhead

Minimal

Full schema per call

Output format

Structured through skills

Agent picks parameters freely

Credits

Same account, same rates

Same account, same rates

Unlimited models

No

No

API key required

No

No

How do I set up CLI + Skills?

  1. Install skills: run npx skills add higgsfield-ai/skills. This pulls three skills: generate, soul, and product-photoshoot.

  2. Sign in: run higgsfield auth login. No API key needed.

  3. Generate: in your coding agent, call /higgsfield:generate or say "Generate an image with Higgsfield." The agent picks the skill and runs it.

Compatible with Claude Code, Cursor, Codex, and other agents that support Skills or MCP.

What are the three Skills?

  • generate, image and video generation from any Higgsfield model (Soul, Nano Banana Pro for images; Seedance, Kling, Veo for video), up to 4K. The agent selects a model automatically, or you can specify one.

  • soul, train a Soul character from photos. The skill handles upload, training, and naming. See How do I create and use a Soul ID character?

  • product-photoshoot, turn a product link or images into studio-quality product shots and video ads. Works for physical products and apps.

Image presets:

What are the three Skills?

Preset

What it generates

product_shot

Product on a neutral, studio, or catalog background

lifestyle_scene

Product in a real-world environment

closeup_product_with_person

Tight crop with hands or partial face and product

pinterest_pin

Vertical 2:3 Pinterest-native format

hero_banner

Wide-format banner for website, email, or campaign

social_carousel

3 to 10 connected slides for Instagram or LinkedIn

ad_creative_pack

Coordinated set of static ad variations

virtual_model_tryout

Product worn or used by an AI-rendered model

conceptual_product

Surreal, CGI-style, or levitating product

restyle

Transform an existing image's aesthetic

Video presets: UGC, Tutorial, Unboxing, Hyper Motion, Product Review, TV Spot, Wild Card, UGC Virtual Try On, Pro Virtual Try On.

Got any questions left?

Two separate systems with the same name. CLI Skills are installed via npx skills add higgsfield-ai/skills for coding agents. Supercomputer Skills are built-in functions called with "/" inside the Supercomputer chat. They are not interchangeable.

No. Use MCP for chat agents like Claude web, and CLI for coding agents like Claude Code or Cursor.

All Higgsfield models. The agent selects automatically unless you specify one.

Update to the latest CLI version by re-running npx skills add higgsfield-ai/skills. Earlier versions had an issue with passing references; current versions pass all references, including images, correctly.

Run npx skills add higgsfield-ai/skills again to pull the latest version.

In your Assets on higgsfield.ai, the same as web and MCP generations.

Share Feedback

Was this article helpful?

by Higgsfield

Share article