fx (by Vercel) is a coding agent harness and CLI written in Zig, optimized for research and embeddability as part of larger systems.
Why it matters
overview
fx (by Vercel) is a coding agent harness and CLI tool developed by Vercel Labs that enables terminal-first developers and researchers to interact with AI models for coding tasks. It is an open-source, native tool written in Zig, designed for minimalism, performance, and embeddability within larger systems or agent infrastructure.
features
fx (by Vercel) provides a minimalist, high-performance environment for AI coding agents, focusing on core functionalities essential for development workflows and research. Its design prioritizes a small footprint and extensibility.
/permissions remember and /permissions revoke commands.--prompt-permissions flag for fx ask runs.use cases
fx (by Vercel) is primarily designed for developers and researchers who require a lightweight, high-performance, and embeddable AI coding agent. Its architecture supports a range of applications from individual development to large-scale system integration.
how to use
fx (by Vercel) operates as a command-line interface (CLI) tool, allowing users to interact with AI models to perform coding tasks directly from their terminal. It enables inspection, modification, and execution of code and shell commands.
fx ask for single, headless requests, such as explaining repository changes, with optional --prompt-permissions.pricing
fx (by Vercel) operates on a freemium model, offering an open-source core with potential for additional paid services or enterprise features. Specific pricing tiers for advanced functionalities beyond the open-source offering are not publicly detailed as of August 2026.
Similar Tools
fx (by Vercel) distinguishes itself in the AI agent landscape through its focus on minimalism, native execution, and embeddability, contrasting with more general-purpose LLM tools or heavier IDE-integrated solutions.
Ollama allows you to run large language models locally on your machine with a simple command-line interface.
While fx is designed as a coding agent harness, Ollama focuses purely on providing an easy-to-use runtime for local LLM inference. You would typically use Ollama as a backend for the models an agent uses, requiring you to build the agent's orchestration logic yourself, whereas fx aims to provide that harness directly.
LocalAI provides a self-hosted, open-source alternative to OpenAI's API, allowing you to run various AI models locally with an OpenAI-compatible API.
LocalAI offers a broader range of models and an API compatibility layer for local inference, managed via CLI, which can be heavier than fx's minimalist approach. fx is a lightweight harness specifically for agents, while LocalAI provides a full local API server for models that agents can then consume.
LiteLLM provides a unified API for over 100 large language models, including local ones, simplifying calls and offering features like fallbacks and retries.
LiteLLM focuses on abstracting away different LLM APIs, making it easier to switch between models and providers, and can be used via CLI or as a library. Unlike fx, which is an agent harness, LiteLLM primarily handles the interaction with the LLM itself, meaning you would still need to implement the agent's specific logic and orchestration around it.
llm is a powerful command-line tool for interacting with and managing various large language models, including local ones, through a flexible plugin architecture.
llm is a general-purpose CLI for direct interaction and management of LLMs, offering extensibility through plugins. fx is designed as a more specialized 'harness' for coding agents, emphasizing embeddability and performance for agent execution rather than general LLM interaction and management. Trade-off: llm is more versatile for general LLM tasks, while fx is tailored for agent orchestration.
More on Stork
Other tools in this category, matched by shared tags