POV: You are staring blankly at your monitor while your coding agent loops on a task, frantically adding optional chaining to an async function because it tried to read a property from a null value.
The agent is caught in the loop it runs over and over. It writes code. It runs code. It reads the error. It tries to fix it.
Except the language it is writing allows nulls and implicit exceptions. Every time the schema shifts slightly, the whole thing crashes at runtime.
I am Jake Burwick, a high school student. I built Vow, an open source programming language under the MIT license designed specifically for AI agents to write. The source code is public at github.com/JakeBurwick/vow.
Vow has no nulls and no exceptions.
If a function can fail, it returns a Result value. An agent cannot just pretend failure does not exist. It must handle the Result value using match or the ? operator, or the code will not build.
fn parsePrice(text) { let n = num(text)? if n < 0 { return err("can't be negative") } return ok(n) } The ? operator unpacks the successful value or returns the error straight up the call stack.
When an agent hits a bug, Vow gives it feedback designed for automated repair. Every Vow error says what is wrong, where, and the likely fix, written as repair instructions an agent can act on without a human.
x TypeError: can't add a num and a str -> line 6, col 10 6 | return "Total: " + sum ^^^^^^^^^^^^^^^ Likely fix: convert the number to text first: str(sum) The whole language, including the lexer, parser, checker, interpreter, and compiler, sits at about 6,400 lines of JavaScript. It compiles to JavaScript. A recursive fib(32) runs in 0.09 seconds, while Python takes 0.14 seconds, measured on October 7, 2026.
Every function in Vow also declares whether it touches the network, files, or environment using uses net, uses fs, or uses env. Running vow --sandbox checks those capabilities before anything runs, and refuses any program that reaches beyond your rules.
You can install it globally with npm:
npm install -g github:JakeBurwick/vow If you use Claude Code, add the MCP server so the agent gets the vow_guide, vow_check, and vow_run tools directly:
claude mcp add vow -- vow-mcp To point your coding agent at Vow, paste a one-paragraph prompt from the site into Claude Code, Codex, Cursor, or any other agent. The site also provides an llms.txt and an AGENTS.md file you can drop into your project.
Head over to https://vow-lang.vercel.app to explore the playground, tutorial, documentation, and the Errors page that displays five bugs side by side in Vow, TypeScript, and Python.
