Bootstrapping Coding Agents: The Specification Is the Program

TL;DR

This paper demonstrates that a coding agent can bootstrap itself by re-implementing a specification from scratch, emphasizing the specification as the stable, primary artifact over the implementation.

Contribution

It introduces a method where an agent re-implements a specification from scratch, showcasing the bootstrap process in AI coding agents inspired by compiler and Lisp concepts.

Findings

The agent successfully re-implements the specification from scratch.

The process confirms the specification's role as the stable record.

Implementation can be regenerated from the specification at any time.

Abstract

A coding agent can bootstrap itself. Starting from a 926-word specification and a first implementation produced by an existing agent (Claude Code), a newly generated agent re-implements the same specification correctly from scratch. This reproduces, in the domain of AI coding agents, the classical bootstrap sequence known from compiler construction, and instantiates the meta-circular property known from Lisp. The result carries a practical implication: the specification, not the implementation, is the stable artifact of record. Improving an agent means improving its specification; the implementation is, in principle, regenerable at any time.