fx is Vercel's tiny, open-source coding agent built to get out of your way. Written in Zig and shipped as a ~6MB native binary, it starts almost instantly while keeping memory and context overhead low. Use it with local or cloud models, extend it with skills, plugins and MCPs, or embed it into your own agent infrastructure. Small by design, so the model gets more room to work.
Vercel just released a coding agent, and it's tiny.
fx is an open-source coding agent written in Zig and packed into a ~6MB binary. It starts almost instantly, uses very little memory, and feels more like a Unix tool than another full IDE inside your terminal.
What caught my attention is that the minimalism isn't just about binary size. fx also keeps the system prompt and tool surface small, so less of your context window is spent on the harness itself.
It's model agnostic, works with local or cloud models, supports Wasm, and can be extended with skills, plugins and MCPs. Vercel also designed it to be embeddable, which makes the idea of running many lightweight coding agents particularly interesting.
It's still very early and experimental, but I like the direction: instead of making the coding agent more complicated, make the harness disappear.
For a while now I've been trying to figure out whether to use Google ADK / Genkit or even Vercel AI SDK for smaller agents. fx seems to be a great candidate to actually try embedding.
It seems I will need to wait for support of LLMs running outside of the Vercel ecosystem since we don't use the Vercel Gateway?
The idea of giving the model more room by reducing tool overhead is interesting. Curious to see what developers build with this.
About fx (by Vercel) on Product Hunt
“Vercel's tiny, open-source coding agent”
fx (by Vercel) launched on Product Hunt on August 21st, 2026 and earned 149 upvotes and 4 comments, placing #4 on the daily leaderboard. fx is Vercel's tiny, open-source coding agent built to get out of your way. Written in Zig and shipped as a ~6MB native binary, it starts almost instantly while keeping memory and context overhead low. Use it with local or cloud models, extend it with skills, plugins and MCPs, or embed it into your own agent infrastructure. Small by design, so the model gets more room to work.
fx (by Vercel) was featured in Developer Tools (517.9k followers), Artificial Intelligence (476.5k followers) and GitHub (41.4k followers) on Product Hunt. Together, these topics include over 222.7k products, making this a competitive space to launch in.
Who hunted fx (by Vercel)?
fx (by Vercel) was hunted by Justin Jincaid. A “hunter” on Product Hunt is the community member who submits a product to the platform — uploading the images, the link, and tagging the makers behind it. Hunters typically write the first comment explaining why a product is worth attention, and their followers are notified the moment they post. Around 79% of featured launches on Product Hunt are self-hunted by their makers, but a well-known hunter still acts as a signal of quality to the rest of the community. See the full all-time top hunters leaderboard to discover who is shaping the Product Hunt ecosystem.
Want to see how fx (by Vercel) stacked up against nearby launches in real time? Check out the live launch dashboard for upvote speed charts, proximity comparisons, and more analytics.
Justin Jincaid
Vercel just released a coding agent, and it's tiny.
fx is an open-source coding agent written in Zig and packed into a ~6MB binary. It starts almost instantly, uses very little memory, and feels more like a Unix tool than another full IDE inside your terminal.
What caught my attention is that the minimalism isn't just about binary size. fx also keeps the system prompt and tool surface small, so less of your context window is spent on the harness itself.
It's model agnostic, works with local or cloud models, supports Wasm, and can be extended with skills, plugins and MCPs. Vercel also designed it to be embeddable, which makes the idea of running many lightweight coding agents particularly interesting.
It's still very early and experimental, but I like the direction: instead of making the coding agent more complicated, make the harness disappear.