Ready to Build: The Verified Rehydration Agent as an Execution-Neutral Specification, Verified Across Five Paradigms

Abstract

We present and demonstrate a property of software specifications that we term execution-neutral buildability. A normative specification has this property when it fixes the system's required behavior and makes any implementation mechanisms it leaves open explicit and bounded by stated obligations. Implementing it is projection — rendering its fixed normative constructs in a target paradigm — rather than designing their operative behavior. The specification's own conformance suite defines the acceptance criteria, including declarative checks that require deployment evidence rather than local execution. The property is defined independently of any particular paradigm: the specification prescribes no host language, runtime, or concurrency architecture, while implementations remain subject to its stated behavioral and deployment obligations. Storage and persistence technologies are likewise unprescribed; the specification nevertheless sets deployment obligations for stable snapshots, anti-rollback protection, local policy compatibility, single-writer enforcement, and readmission authority. We demonstrate the property on a single, final, manifest-verified artifact, the Verified Rehydration Agent pack (AGT-VRA-001), available free of charge under its included license, and report the central cross-provider reconstruction result: five paradigm-isolated contexts built the specified agent logic from the completed pack alone in an AI-provider environment different from the one used to construct its canonical files — in Python, Rust, Prolog, C, and Bash — and ran the pack's 29 conformance records live, on local toolchains attested by a live handshake rather than a static trace. Compared across the five per-paradigm runs, all 29 records reached the same outcome in every run: 27 PASS, 2 SKIP (PT-028 V_COMPAT and PT-029 V_WRITER_LOCK: declarative deployment checks outside the local run surface), 0 divergent. We argue, but do not claim to have proven, that this convergence is evidence of the derivability and self-sufficiency of the specification — not a certification, not a quality score, and not a set of reference implementations to be followed. Readers can verify artifact integrity by checking the shipped hashes against the Manifest, and reproduce the convergence result by rebuilding the five baselines from source with the documented flags and rerunning the 29-record suite. The pack's self-audit runbook guides package verification; per-paradigm rebuild instructions are provided with the corresponding baselines. Separately, Section 5 reports five further target paradigms (Common Lisp, Brainfuck, spreadsheet formulas, an Analytical Engine-style program, and AWK): the first four participate in a six-rendering sweep (with Python and Rust comparators) of the declared 162-input state-transition projection, while AWK and spreadsheet formulas report full-suite results. We are explicit about limits: the reproducible finding is cross-paradigm convergence; we interpret it as evidence of structural derivability, not semantic completeness or domain fidelity, and the authoring methodology that produces such specifications is out of scope of this paper. Use of AI tools. The author constructed the pack's seven canonical files, including the conformance tests, reviewed them manually, and designed and calibrated the harnesses for the reconstruction campaign and the supplementary derivations. Large language models were used openly, as tools to accelerate research and production, with manual review: they assisted the drafting of the pack's documents and the revision of this manuscript and, by design, built the reconstructions and the supplementary derivations that are the object of the demonstration. No tool is an author; the author is accountable for the content. What matters is not which tools touched the documents but whether the artifact can be verified. The claims in this paper are limited to what a reader can check with the shipped hashes, the runnable conformance suite, and the reproducible outcome vectors.