Meta releases Muse Code, an AI helper for large software projects

In short: Meta has launched Muse Code in beta, an AI tool meant to help programmers work on large and complex software projects.

What happened

Meta announced Muse Code, a “terminal” coding agent, which means it runs from a text-based command screen many developers already use. Meta CEO Mark Zuckerberg said it can handle full software tasks across large “repos” (short for repositories, which are like big folders that store a project’s code).

Zuckerberg said Muse Code can plan changes, write code, and validate results, which means it can run checks to see if the new code works as expected. Meta says the tool can be installed with a single command.

Muse Code is powered by Meta’s earlier coding model called Muse Spark. For larger jobs, it can start multiple “sub-agents” (like assigning parts of a big group project to several helpers at the same time). Zuckerberg said these helpers work in separate, isolated work areas so your main copy of the code is not changed directly, and in testing it built six game features at once without conflicting changes.

Meta is positioning Muse Code against similar tools from OpenAI and Anthropic. Meta AI chief Alexandr Wang told The Wall Street Journal that the product could be a strong option for many uses, especially on cost.

Why it matters

Many apps and websites run on huge piles of code built over years, and making changes can be slow and error-prone. Tools like Muse Code aim to reduce that busywork and help teams ship updates faster, which can affect how quickly products you use get fixes and new features.

Source: TechCrunch AI