Coding & app building

fx by Vercel

By fx.sh

fx by Vercel is a strong fit for scripting or embedding a lightweight coding agent into your own terminal or developer workflow, with a profile optimized for advanced users who value medium ease-of-use and medium output quality.

Best for: Scripting or embedding a lightweight coding agent into your own terminal or developer workflow

What it is

Vercel Labs' tiny, open-source (Apache-2.0) embeddable coding-agent CLI written in Zig — a single ~8 MB native binary with a Unix shell-style interface, made to be scripted, embedded (WebAssembly, JavaScript APIs, Agent Client Protocol, MCP) and run model-agnostically against local models, the Vercel AI Gateway, or your existing ChatGPT (Codex) / Grok subscription.

In Choosely terms, this sits in the coding & app building lane and is commonly selected for scripting or embedding a lightweight coding agent into your own terminal or developer workflow and running a model-agnostic coding agent via the vercel ai gateway or a chatgpt/grok subscription.

Pricing

fx itself is free and open source (Apache-2.0); there is no charge for the CLI. Model inference is billed separately by whichever provider you connect — Vercel AI Gateway credits, or your existing ChatGPT (Codex) or Grok subscription. No permanent free provider access is bundled.

Basis: Open SourceConfidence: VerifiedLast checked: August 2026

Why people pick it vs where it falls short

Why people pick it

  • Tiny native binary (~7.8 MiB, built in Zig) with fast cold starts; installs via a single setup script and runs locally
  • Model- and provider-agnostic: Vercel AI Gateway (API key), or OpenAI Codex / xAI Grok via subscription OAuth, plus local models
  • Embeddable and scriptable — WebAssembly (createFxAgent / createFxTerminal), JavaScript APIs, Agent Client Protocol, MCP servers, skills and subagents

Where it falls short

  • Explicitly experimental Vercel Labs project (v0.0.x) with frequent breaking changes — not a stable production tool
  • Not an IDE, a visual app builder (that is Vercel v0), a deployment/hosting platform, or a model gateway itself; you bring your own model/provider and pay those costs separately
  • Terminal-first and, when building from source, needs the Zig toolchain — aimed at developers, not beginners

When it is a strong fit

A strong match when your main priority is scripting or embedding a lightweight coding agent into your own terminal or developer workflow and you need an advanced-friendly starting point.

Useful when your team values medium ease of use and fast execution over heavier setup.

Best when medium quality matters, but you still want a practical workflow rather than a complex implementation track.

How it compares in Choosely terms

  • Speed profile: Fast. This is best when you want momentum from prompt to usable output without heavy process overhead.
  • Ease profile: Medium for Advanced users. You can move quickly even if this is not your full-time specialty.
  • Control profile: High. Expect practical customization, but not an infinite-control architecture.
  • Pricing signal: Open-source. Good for teams balancing capability with cost sensitivity.
Tradeoff: Explicitly experimental Vercel Labs project (v0.0.x) with frequent breaking changes — not a stable production tool.

Best-fit use cases

Practical ways fx by Vercel fits the current Choosely catalog profile.

Lightweight Unix Style Coding Agent Cli I Can Script And Embed In My Own Workflow

Strong lane

Use fx by Vercel for lightweight unix-style coding-agent cli i can script and embed in my own workflow when you want fast execution, medium ease of use, and medium output quality.

Small Model Agnostic Agent Harness For Terminal Automation

Strong lane

Use fx by Vercel for small model-agnostic agent harness for terminal automation when you want fast execution, medium ease of use, and medium output quality.

Embeddable Coding Agent Runtime That Uses Models Through The Vercel AI Gateway

Strong lane

Use fx by Vercel for embeddable coding-agent runtime that uses models through the vercel ai gateway when you want fast execution, medium ease of use, and medium output quality.

A Tiny Native Coding Agent I Can Embed With Webassembly And The Agent Client Protocol

Strong lane

Use fx by Vercel for a tiny native coding agent i can embed with webassembly and the agent client protocol when you want fast execution, medium ease of use, and medium output quality.

Alternatives

Claude Code

Anthropic's coding agent for working across codebases, terminals, fixes, and longer-horizon development tasks.

Choose Claude Code if output quality consistency matters more than raw speed.

OpenAI Codex

Cloud-based software engineering agent platform from OpenAI for delegating coding tasks, reviewing changes, and operating across repository workflows.

Choose OpenAI Codex if output quality consistency matters more than raw speed.

Next step

Install with `curl -fsSL https://fx.sh/setup.sh | bash`, run `fx setup` to connect the Vercel AI Gateway or your ChatGPT/Grok subscription, and try a small scripted task before embedding it via WASM/ACP — pinning a version, since it changes frequently.

Related reads

FAQ

What is fx by Vercel best for?

fx by Vercel is best for scripting or embedding a lightweight coding agent into your own terminal or developer workflow, running a model-agnostic coding agent via the vercel ai gateway or a chatgpt/grok subscription, a minimal, native (wasm-embeddable) alternative to heavyweight ide coding assistants.

Is fx by Vercel beginner-friendly?

This catalog profile lists fx by Vercel at advanced skill level with medium ease of use.

What should I watch out for before choosing fx by Vercel?

Explicitly experimental Vercel Labs project (v0.0.x) with frequent breaking changes — not a stable production tool