🤖 AI Summary
This work proposes a novel approach to endow code-generating agents with self-guided and regenerative capabilities, thereby eliminating dependence on an initial implementation. Inspired by compiler bootstrapping, the method leverages a concise 926-word specification document to drive a large language model–based agent to correctly reconstruct its own implementation from scratch. This study presents the first demonstration of a bootstrapping mechanism in AI coding agents, validating the principle that “the specification is the program” and establishing a high-quality specification as the sole stable artifact. Experimental results show that, given a complete specification, the implementation can be accurately regenerated at any time, underscoring the critical role of specification-driven development and meta-circular architectures in autonomous agent systems.
📝 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.