chimera skills-evolve

Reflectively evolve a skill's prompt template against graded instances (GEPA). Each instance is a `{input, expect}` pair; the (simple, honest) scorer gives 1.0 when the produced output contains the `expect` substring, else 0.0. GEPA reflects on a failing case to rewrite the template and keeps a Pareto frontier of candidates. Dry-run by default: the improved skill is only written back to the store with ``--apply``, and only if it beats the seed.

Аргументы

NAMEstrобязательный
Name of the learned skill whose prompt to GEPA-evolve.

Параметры

--instancesstrобязательный
JSON file: a list of {"input": {...}, "expect": "substring"}.
--budgetintпо умолчанию: 20
Rollout budget (evaluations across the search).
--modelstr
Model slug for the executor + reflector.
--applyboolean
Save the improved skill (default: dry-run).