mlsanityData to 7 Oct 2026

Bootstrapping Coding Agents: The Specification Is the Program

arXiv 2603.17399 ↗First posted 18 Mar 2026cs.SE cs.LGcited by 1

Not on the map. Why?

Abstract (arXiv)

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.

Key facts. Cited by 1 paper in the map (data through 7 Oct 2026). Builds on 4 of its 13 references that are in the map. First posted on arXiv 18 Mar 2026.

Builds on

4 of 13 references.

References · 4

  1. 2023SWE-bench: Can Language Models Resolve Real-World GitHub Issues?cited by 3,193
  2. 2026Spec-Driven Development:From Code to Contract in the Age of AI Coding Assistantscited by 14
  3. 2024Galapagos: Automated N-Version Programming with LLMscited by 5
  4. 2022Debootstrapping without Archeology: Stacked Implementations in Camlbootcited by 1

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026From Prompt to Harness: Coderlet from Scratchcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026BootstrapAgent: Distilling Repository Setup into Reusable Agent Knowledge
  2. 2026Position: Coding Benchmarks Are Misaligned with Agentic Software Engineering
  3. 2025On the Use of Agentic Coding Manifests: An Empirical Study of Claude Code
  4. 2026Specification-first convergence with an AI coding agent: a case study of dismantling a core architectural invariant across 189 files in a 717k-line codebase with no test oracle and no human code review
  5. 2026What makes a harness a harness: necessary and sufficient conditions for an agent harness
  6. 2026Agent Behavior as Code: Efficient and Robust LLM Agents with Programmatic Specifications
  7. 2026Verus-SpecGym: An Agentic Environment for Evaluating Specification Autoformalization
  8. 2026Are We All Using Agents the Same Way? An Empirical Study of Core and Peripheral Developers Use of Coding Agents