verify
Installation
SKILL.md
Verify
Sample N isolated trajectories, then rank them (Kwok et al., arXiv:2607.05391). One score token is an LM judge, not this method.
Flags
| Flag | Default | Effect |
|---|---|---|
--candidates <n> |
3 |
Pool size N. Not inferred from the task |
--evals <k> |
2 |
Repeats K; swap A/B on every odd pass |
--pivots <k> |
2 |
PPT pivots when N>3; clamp to [1, N] |
--max-rounds <n> |
0 |
Extra generate or revise cycles after select |
--criteria <path> |
auto | Criteria file; else write 2–4 inline |
--track |
off | Progress only; skip generate and select |
A supplied list overrides --candidates. N is a budget, not a computed count.