Legacy DSH skill report
DSH runtime guide
null / unstable. Sandbox-awareness and true-resume are controls; runtime-upgrade behavior is candidate-unstable, not a guarantee.
6/6 bothsandbox awareness
6/6 bothtrue-resume
6/6 → 5/6runtime-upgrade
5/6 → 3/6guide without
Disabled: No stable positive signal; kept for rework, hidden from default skill discovery. The skill is kept in the repository as
SKILL.md.disabled.Do not claim a stable runtime-guide improvement; two cases are controls/null and one is unstable.
Documented comparisons
legacy evidence, not sealed records| Case / batch | With skill | Without skill | Reading |
|---|---|---|---|
| sandbox-awareness paired | 6/6 | 6/6 | control, no correctness or call benefit |
| true-resume at target cutoff | 6/6 | 6/6 | reproducible null result |
| runtime-upgrade at 100 | 6/6 then 5/6 | 5/6 then 3/6 | candidate-unstable |
| upgrade-protocol explicit refusal | 6/6 then 5/6 | not scored | incomplete second run; not a guarantee |
Source and limitations
These numbers come from benchmarks/evaluation/TASK_LEVEL.md and benchmarks/evaluation/FINAL_REPORT.md. Raw trial records were not retained as sealed records in this rebuild; counts here are documented legacy results.
No cross-skill or cross-model aggregate is computed. The attribution report is the only page in this site backed by the current sealed evidence set.