Companion to Companion Executable Code Repositories

A rerun protocol, a result record, two checklists, a layout and a metadata profile, none of them tested

This companion implements the Intake draft of the program "Companion Executable Code Repositories", dated 2026-10-22. The program states the claims. The instruments here are the ones it proposes, as designed, and none has been used yet.

A protocol for rerunning a published paper’s code, the practical test of H1 and H2.
A result record for one rerun.
An author’s checklist for the six parts of a companion repository.
A reviewer’s checklist for rerunning one.
An example layout for a companion repository.
The draft metadata profile, over four existing standards.

Following the checklists does not show a paper is right

The checklists say what the program recommends a companion repository hold, in "A companion repository holds six parts". Following them doesn’t show that a paper is right. It shows that a person other than the author can rerun the analysis and compare the numbers. Whether that catches errors that matter is H4, and it is untested.

A change to what an instrument measures or asks is made in the program first, and then carried here.