Appendices

Six technical appendices, for readers who want to check what the article claims. Each is optional, and each can be read on its own.

Appendix What it shows
A How long would the configuration tool have taken by hand? An estimate of the effort and cost of building the largest component by hand, with its method and which way each assumption leans
B Do the AI's tests catch anything? Mutation testing on the device and the configuration tool, the defect every test missed, and what was fixed
C What does the hash chain protect against? The history of every change, opened down to its database triggers: what it protects, and against whom it does not yet
D How was the agent steered? Which model did what, how large a story's brief became, what the reviewers found, and what only a person found
E What do the quality numbers say? Dependencies and their known advisories, static analysis, build and response times
F What does the QMS plugin do? The published plugin that puts the quality system in front of every workflow, with its tests

Appendix A is for anyone weighing the cost. Appendices B to F are written for engineers and for quality and regulatory readers, and follow one pattern: what was checked, what was found, what was done about it, and what is proposed next.

Each opens with its status, the date of the snapshot it describes and what is still open. Numbers in the form OPEN-nnn are entries in the project's register of open items, part of its quality system (chapter 06).