Problem

KnowledgePoint model has two validation issues when used with LLM-extracted knowledge points:

1. document_id is required — LLM extraction doesn't produce this field, causing Field required [type=missing] error
2. prerequisites expects UUID list — LLM outputs prerequisite titles as strings (e.g., ["智能体的定义"]), causing uuid_parsing error

Steps to Reproduce

1. Extract knowledge points from a document using KnowledgeExtractorAgent
2. Save results to knowledge_points.json
3. Load and validate with KnowledgePoint.model_validate(kp)
4. Pydantic raises validation errors

Expected

The model should be flexible enough to handle LLM-generated data.

Suggested Fix

1. Make document_id optional:

document_id: UUID | None = None

2. Add a validator to coerce string prerequisites to UUIDs:

@field_validator("prerequisites", mode="before")
@classmethod
def coerce_prerequisites(cls, v: list) -> list[UUID]:
    result = []
    for item in v:
        if isinstance(item, UUID):
            result.append(item)
        elif isinstance(item, str):
            try:
                result.append(UUID(item))
            except ValueError:
                result.append(uuid5(NAMESPACE_DNS, item))
    return result

Environment

• Python 3.14, Windows 11, latest main branch