Skip to main content

agrag.agents.verification.VerificationResult

Bases: BaseModel

The verifier's structured verdict on the researcher's findings.

The verifier returns this through its response_format, so the planner reads a typed verdict out of the task tool's result instead of parsing free-form prose.

Attributes:

  • reasoning (str) – What the independent per-sub-question checks found, written before the verdict is decided.
  • status (Literal['PASS', 'INSUFFICIENT', 'CONTRADICTORY']) – The overall verdict. PASS means every sub-question has supporting evidence and nothing contradicts; a re-delegation after this verdict is not a retry. INSUFFICIENT means one or more sub-questions lack supporting evidence, listed in missing_evidence. CONTRADICTORY means cited evidence conflicts, which re-researching cannot resolve.
  • missing_evidence (list[str]) – The sub-questions and evidence gaps to close, filled when the status is INSUFFICIENT.

missing_evidence​

missing_evidence: list[str] = Field(default_factory=list)

reasoning​

reasoning: str

status​

status: Literal['PASS', 'INSUFFICIENT', 'CONTRADICTORY']