Appendices

All measurements were read-only against local git histories, local agent session logs, and the GitHub API. Appendices G and H disconfirmed the earlier task-class rework-map moat and are the reason autonomy is granted per contract.

1. A. Local git backtest
2. B. Public evidence
3. C. Negative-class census
4. D. Session to commit linkage
5. E. Written demand signals
6. F. Restricted rework metric
7. G. Public repo rework map
8. H. Split-half stability