Class EvolutionSuggestion

openspace/skill_engine/types.py:198–248

One evolution action suggested by the analysis LLM. ``target_skill_ids`` lists the parent skill(s) this action targets using **true skill_id** values (e.g. ``weather__imp_a1b2c3d4``): - FIX: exactly 1 parent (the skill to repair in-place) - DERIVED: 1+ parents (single parent → e

@dataclass
class EvolutionSuggestion:
    """One evolution action suggested by the analysis LLM.

    ``target_skill_ids`` lists the parent skill(s) this action targets
    using **true skill_id** values (e.g. ``weather__imp_a1b2c3d4``):
      - FIX: exactly 1 parent (the skill to repair in-place)
      - DERIVED: 1+ parents (single parent → enhance; multi → merge/fuse)
      - CAPTURED: empty list (brand-new skill, no parents)
    """

    evolution_type: EvolutionType
    target_skill_ids: List[str] = field(default_factory=list)  # True skill_id(s)
    category: Optional[SkillCategory] = None                   # Desired category of the result
    direction: str = ""                                        # Free-text: what to evolve / capture

    @property
    def target_skill_id(self) -> str:
        """Primary (or only) target skill_id.  Empty string if none."""
        return self.target_skill_ids[0] if self.target_skill_ids else ""

    def to_dict(self) -> Dict[str, Any]:
        return {
            "type": self.evolution_type.value,
            "target_skills": self.target_skill_ids,
            # Keep legacy singular key for backward compat with stored analyses
            "target_skill": self.target_skill_id,
            "category": self.category.value if self.category else None,
            "direction": self.direction,
        }

    @classmethod
    def from_dict(cls, data: Dict[str, Any]) -> "EvolutionSuggestion":
        cat = None
        if data.get("category"):
            try:
                cat = SkillCategory(data["category"])
            except ValueError:
                pass
        # Support both new list format and legacy single-string format
        raw_targets = data.get("target_skills")
        if isinstance(raw_targets, list):
            targets = [t for t in raw_targets if t]
        else:
            legacy = data.get("target_skill", "")
            targets = [legacy] if legacy else []
        return cls(
            evolution_type=EvolutionType(data["type"]),
            target_skill_ids=targets,
            category=cat,
            direction=data.get("direction", ""),
        )


# Task-level execution analysis (1 per task)